Hardware Announcement
May 7, 1998
Announcement Letter Number: 198-115
Overview
IBM is introducing 16 new Generation 5 S/390 (R) Parallel Enterprise Server models. Fifteen general purpose and one Coupling Facility model exploit the latest IBM Complementary Metal Oxide Semiconductor (CMOS) technology yielding up to 2x the performance of the corresponding G4 models. The 9672 R06 is a more powerful Coupling Facility which can be upgraded to other G5 models. 9674 C04 and C05 CF's can upgrade to 9672 R06's. All G5's offer substantial performance and functional and feature options over previous generations of CMOS and bipolar Systems.
New functions and features such as IEEE Floating Point, TDES cryptography with dual path cryptographic coprocessors, FIbre CONnection channels (FICON (TM)), Internal Coupling channels, Integrated Cluster Bus coupling connections and Fast Ethernet connections are available on G5 models. G5's design excellence balances processors, memory and I/O connectivity, optimizing them for Enterprise computing. Memory and I/O bandwidth has increased over 100% from previous CMOS servers. New architecture, tailored to TCP/IP performance improvements, makes G5 servers extremely powerful and scalable commercial Web Servers. In addition to the new optional Fibre Channel, G5 Servers will also support Geoplexes and new Open Systems Adapter-2 technology. Parallel Sysplex (R) cluster technologies also have added flexibility, performance and connectivity.
The new models provide the Parallel Sysplex customer more Single Systems Image (SSI) capacity, continuous availability, scalability and performance, while providing the single system customer with the upgradeability, performance and functionality to increase existing workload capability, and provide increased capacity for the new workloads every enterprise has witnessed. G5 has significant improvements for decimal instructions compared to G4. The S/390 G5 Parallel Enterprise Server models also provide increased performance and throughput for customers who have substantial floating point arithmetic calculations and other types of numerically intensive workloads. Furthermore, the S/390 G5 Parallel Enterprise Server models also offer new instructions to further enable C/C++, and Java (TM).
Intended Customers
The new S/390 G5 Parallel Enterprise Server models' customer segment includes the existing S/390 processor inventory up through the ES/9000 (R) 9021 711-based processors, upgrades from 9672 model R2, R3, G3 and G4 servers, and 9674 Model C04 & C05 coupling facilities as well as equivalent Plug Compatible Machine (PCM) inventory.
Both existing traditional and new workloads will benefit from the G5 server's robust performance, scalability and improved availability.
S/390 Satisfaction Guarantee
To translate the quality of S/390 processors into a stronger commitment to our customers, IBM is offering a Satisfaction Guarantee. Refer to the Description section for details.
------------------------------
This announcement is provided for your information only. For additional information, contact your IBM representative, call 800-IBM-4YOU, or visit the IBM home page at: http://www.ibm.com
S/390 G5 Satisfaction Guarantee
To translate the quality of the S/390 processors into a stronger commitment to our customers, IBM is offering the following S/390 Satisfaction Guarantee in countries where it is applicable:
If you, the original user of an IBM S/390 Parallel Enterprise Generation 5 Server, or S/390 Generation 5 Coupling Facility are dissatisfied with its quality or reliability, IBM will replace your equipment with the same machine model or model upgrade, when requested in writing by you within one year of its initial installation date. The replaced equipment must be transferred to IBM with clear title, free of liens and encumbrances, and in an unaltered condition. This guarantee applies only in the country of purchase, does not apply to replacement equipment, and does not cover damage caused by alterations, misuse, accident, unsuitable physical or operating environment, improper maintenance, or improper installation by a third party.
IBM may change the terms of this guarantee at any time. However, any change would not be retroactive.
Planned Availability Dates
Other Availability Announcements
New S/390 G5 Parallel Enterprise Servers-9672 Upgrades
Model Conversions
Model conversions will be announced at Product Announcement second quarter 1998.
Performance Information
LSPR Measurements will be available September 1998.
Preliminary performance measurements on a limited model set confirm up to 2X performance improvement, compared to G4 models.
S/390 Parallel Enterprise Generation 5 Server -- Design Excellence
Enhanced Generation 5 CMOS Bi-nodal Processor Sets: G5 processors utilize CMOS 6X technology to pack approximately 4 million circuits on a 14.6mm chip. G5 processor sets have 6 Processor Units supported by a Storage Control chip, 4 Storage Data chips and 2 Memory Bus Adapter chips. The largest models have 2 sets of 6 PUs each which can support up to a 10 Way Multi-Processor and 2 System-Assist Processors (SAPs) -- dedicated I/O processors.
The chip count on the Generation 5 MultiChip Module (MCM) was reduced from Generation 4 from 30 to 29. Securely enclosed within the module are two dual path Crypto Chips and the System Clock (Clk) chip. These 3 chips, along with the 13 (1 processor set) or 26 chips (2 processor sets) are mounted using IBM's Controlled Collapse Chip Connection (C4) technology on advanced ceramic substrates.
The G5 processor set design eliminates bottlenecks and contention of previous CMOS and BiPolar designs. G5 has a new System "nest" structure that uses three system busses an improvement over G4's single system bus that handles address, data and controls for memory fetches. The cache structure of G5 is also improved. G5 has 256 KB of Level 1 cache on each PU. G5's large on chip L1 cache increases the L1 cache hit success ratio for most workloads. Backing G5's larger L1's is a larger and improved Level 2 cache located on the SD chips. The Processor sets have 4 MB of shared Level 2 cache, 1 MB on each of the SD chips in a processor set. L2 cache is shared on 6 PU's in two G5 sets, an improvement over the G4 design of sharing L2 across 3 PU's in 4 sets. The G5 System design having up to 8 Mbytes of L2 cache (2 sets) eliminates the need for Bus Switching Network (BSN) chips that comprised an additional cache level (sometimes called an L 2.5) that was part of CMOS G3-G4 System designs. The G5 memory line size has also been increased to 256 bytes to allow additional prefetching. G5's cache access improvements, contention elimination and cycle time improvements (approximately 20% over G4 models) combine to increase performance over G4 and 711 based models (refer to previous section -- Performance Advantages).
The use of a Branch History Table (BHT) enables G5s to have more throughput than G4s. CMOS integration and the integration of processor elements in the single PU chip, with a larger balanced cache structure enables G5 Systems to have more throughput than 9021 711 based processors. G5 processors have much less variability of throughput for different workloads than G4 processors. The streamlined balanced design of G5 processor sets significantly reduces queing effects that limits BiPolar or CMOS (high end) processor performance.
Generation 5 Processor sets can utilize up to 24 Self Timed Interconnect (STI) busses to gather or send data. The STI bandwidth of 333 MB/sec bidirectional allows data to be moved to and from the System in a fast and efficient manner. G5's can have up to 24 GB of storage, giving customers the ability of having large amounts of central memory (Level 3) to support up to 15 LPARS.
Reduced Cost of Computing
The new S/390 G5 Parallel Enterprise Server models continue to enable significant savings with respect to energy, facilities overhead, and maintenance costs when compared to older technology processors. Customers who migrate from ES/9000 or comparable non-IBM systems can see up to 97% savings in energy costs (including power and cooling), up to 65% in service and maintenance costs, and more than 91% in floor space requirements. Low energy requirements leads to economies in facilities infrastructure. This may avoid machine room expansion, additional power distribution retrofits, and additional heat exchanger burden. The universal electrical primary source capability of IBM's new servers accommodate direct attachment to any electrical utility worldwide.
Increased Server Storage Capability to 24 GB: IBM leads the industry by offering up to 24 GBs of server storage capability on the new S/390 G5 Servers and Coupling Facility R06 model. This enhances the ability to balance S/390 G5 system resources with the performance of the processor sets. In PR/SM (TM) LPAR mode, models with greater than 2 GBs installed storage can configure some or all of the installed storage as central storage. Each individual logical partition is limited to a maximum of 2 GBs of central storage. The sum total of all logical partition central storage can exceed 2 GBs up to the limit of processor storage.
Generation 5 Memory Structure -- Capacity and Additions
Generation 5 models use 1 GB, 2 GB, 3 GB, 4 GB and 6 GB memory cards. Models RA6, R16, RB6, R26, RC6 and RD6 always have 2 cards. Models R36, R46, R56, R66, R76, R86, R96, RX6 and YX6 have 4 cards. Model R06 has either 2 or 4 cards depending on the number of CF engines or ICF's. Model R06 with 1 to 4 ICF's has 2 cards and an R06 with 5 to 10 ICF's has 4 cards. Models R16 and R26 normally have 2 cards.
The maximum memory for any G5 model is achieved using 6 GB cards. Systems with 2 cards have a 12 GB maximum; those with 4 cards have a 24 GB maximum.
Minimum memory configuration is determined by using one half of the smallest memory increment (the 1 GB card) capacity. Systems with 2 cards have a 1 GB minimum and systems with 4 cards have a 2 GB minimum.
Memory granularity has been increased; when a maximum memory option of a physical card is reached, cards must be swapped out for larger capacity cards. LIC upgrades require an IML to initialize the new increment of memory and identify the feature properly in the Vital Product Data.
The following table identifies memory increments and physical boundaries. All values are in GBs.
Total System Memory Capacity of 2 Card Models:| Capacity of | Amount of | (NOTE) | | Physical Card | Memory available | | | (GB) | | | | 1 | 1.0 | | | 1 | 1.5 | | | 1 | 2.0 | (1) | | 2 | 2.5 | | | 2 | 3.0 | | | 2 | 3.5 | | | 2 | 4.0 | (1) | | 3 | 5 | | | 3 | 6 | (1) | | 4 | 7 | | | 4 | 8 | (1) | | 6 | 10 | | | 6 | 12 | (1) | | | | | ------------------------------------------------- | Total System Memory Capacity of 4 Card Models: | | | | | Capacity of | Amount of | (NOTE) | | Physical Card | Memory available | | | (GB) | | | | 1 | 2.0 | | | 1 | 2.5 | | | 1 | 3.0 | | | 1 | 3.5 | | | 1 | 4.0 | (1) | | 2 | 5 | | | 2 | 6 | | | 2 | 7 | | | 2 | 8 | (1) | | 3 | 10 | | | 3 | 12 | (1) | | 4 | 16 | (1) | | 6 | 20 | | | 6 | 24 | (1) |
(1) Boundary point: additional memory capacity requires a hardware replacement.
For stocking and service flexibility a larger memory card may be used to replace a smaller one; Licensed Internal Code (LIC) establishes a logical equivalent. This Generation 5 memory design is new to 9672's (which previously required identical memory cards to be in all positions). Upgrades from 2 card models to 4 card models may have cards of equal physical size if a memory upgrade is required or cards of 2 different sizes if the total system memory is to remain the same.
All Generation 5 models can use the same memory cards. G5 memory cards have an improved design over any previous S/390 memory cards and use the latest fast 64Mb Synchronous DRAMs.
G5 Memory Value and Improved Design: G5 memory scales with the number of processors supported. Models RA6-RD6 are configured with 2 memory cards and hence are known as a 2 bus, or 2 card system. Models R36-YX6 are configured with 4 memory cards and are called 4 bus (or 4 card) systems. Model R06 is configured with 2 cards for configurations with 1-4 ICF's and 4 cards for 5-10 ICF's. Cards have various amounts of memory which can be flexibly used for System memory granularity. Total memory can be ordered in the following GB increments: 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, 6, 7, 8, 10, 12, 16, 20, 24. G5 memory card sizes are 1,2,3,4 and 6 GB. Memory can be upgraded in many cases by microcode. It is only when the maximum size of System memory supported by a given card size is exceeded that an upgrade which swaps out cards for higher capacity cards is required. All memory cards are high performance cards. Each card or Primary Memory Array (PMA) contains 8 CMOS 5S memory adapters, 2 CMOS 6S Redrive chips and 64 MB Synchronous DRAM technology. The minimum amount of memory provided with G5 models is 2 to 4 times more than G4 models.
G5 memory design has removed potential bottlenecks which improves throughput and reduces workload variability when compared to G4. The memory interface now handles 4 independent memory requests, one per address bank on the Processor Memory Array (PMA is a card). This speeds data flow. Also by separating the address/command from the data bus potential contention is reduced. Hierarchical Storage Control with the 4 MB L2 in the System Storage Controller (SC) enhances direct controlled memory operations. The SC is designed to support up to 32 GB of main store. This allows expansion of G5's present memory maximum, 24 GB, when required in the future.
The G5 System has approximately 6 GB/sec, 12 GB/sec or 15 GB/sec of memory bandwidth depending on the system model. Models with one processor set (RA6 -RD6) have 6 GB/sec of memory bandwidth. Models with two processor sets (R36 -- RX6) have 12 GB/sec of memory bandwidth. The YX6 model, which runs at the fastest cycle time, is supported by 15 GB/sec of memory bandwidth. Advanced 64 Mb memory technology, multiple memory paths, innovative processor design, advanced memory redrive chips, and fast cycle time combine to give the G5 memory superior performance. Processor design innovations include non-blocking cross-point switching and a dedicated Move Page engine, which significantly improves Main-to-Expanded paging performance.
In addition to improvements for performance, G5 memory has reliability and availability improvements compared to either CMOS or bipolar processors. These improvements include the data bus being Error Checking and Correcting (ECC) protected. G5 memory also has the ability to delete and relocate L1 and L2 data. G5 Main Store and Expanded Store has ECC, background scrubbing, and dynamic sparing of L3/L4 memory which provides memory capable of supporting the extremely reliable CMOS PU sets.
I/O Infrastructure
The Generation 5 models all have been designed to eliminate present and future I/O constraints to running intensive workloads. Given the new milestone in CMOS Server performance, Parallel Sysplex installations should insure maximum data rate capability by installation of full speed InterSystem Coupling (ICS) fiber optic links (100 Mbytes/second). These links exploit full configuration ability and maximum flexibility in conformance to world-wide telecommunications standards. The ISC link uses Long Wavelength laser (1300 nm.) Single-mode fiber (SMF) and operates at 106.5 Mbit/second. This technology enables parallel sysplex configurations connections from several meters to distances up to 10 km.
G5's I/O infrastructure's bandwidth has been increased by over 100%.
Introducing the S/390 FIbre CONnection (FICON) Channel: FICON brings to the customer a new High-performance I/O channel, optimized for efficiency at high-speed. At the same time, it provides customers with investment protection, system balance, and coexistence with today's parallel, ESCON TM, OSA, and clustering link technologies. This is an evolutionary step for S/390 -- not one of revolution. FICON takes advantage of the ANSI Fibre Channel Standard transport with the introduction of a performance-enhanced S/390 layer.
FICON's high performance can help meet the needs of evolving workloads of network computing, business intelligence, and other new applications. The high performance also supports the improvements in device transfer rates when future FICON-compatible devices are natively-attached to a FICON channel.
Each FICON channel is capable of supporting more than 4000 I/O operations per second, which allows each channel to support the same capacity as up to 8 ESCON channels. The channel link speed is 100 MB per second full duplex compared with 17 MB per second simplex with ESCON links. The full duplex nature of FICON channels permits them to read and write data concurrently on the same link.
FICON channels achieve these capabilities in a manner which also permits intermixing of large and small I/O operations on the same link, a significant enhancement compared with ESCON channels. Configurations can be thought about in a different manner because of the increased I/O rate, increased bandwidth, and multiplexing of mixed workloads.
FICON's high I/O rate and bandwidth permits channel consolidation by increasing total system I/O capacity. This I/O capacity increase permits growth of single images and logical partitions. Up to 12 FICON channel cards are initially supported on G5.
The 9032 Model 005 Director will be the first product to support FICON through the use of a bridge card and is intended to provide investment protection for currently installed ESCON control units. Up to 16 bridge cards will be supported on a single 9032 Model 005, with each card capable of supporting up to 8 concurrent ESCON control unit connections. Control units can be connected to the system using both FICON (through the bridge card) and ESCON, preserving the customer's investment in all ESCON Directors, as well as the investment in the existing fiber infrastructure. In addition, the performance-enhanced S/390 layer preserves the investment in software which creates channel programs, for control units attached through the bridge card and future natively-attached FICON control units.
The new FICON channels are bigger and faster, supporting the characteristics of both existing and evolving workloads as well as the higher-speeds of new and future devices. But FICON is not just a bigger, faster channel. FICON doesn't provide you with more channels, it provides you with better channels that improve system balance while permitting single image or partition growth, and support mixed workloads on the same channel.
With their improved capabilities, the new FICON channels enable easier Server Consolidation, New Application Growth, large Business Intelligence queries, and exploitation of Network Computing.
G5 Parallel Sysplex Improvements
The Generation 5 CMOS Servers offer the most substantial cluster technology enhancements of any IBM S/390 server introduced to date. The G5 Server Internal Coupling Facility (ICF) is the most powerful CF Uni- processor yet and is key to the Parallel Sysplex technology integration strategy. The new 9672 R06 standalone coupling facility is the most is the most powerful CF in terms of uni-processor and CEC capacity, as it supports up to 10 ICF processors. The R06 is directly upgradeable to 9672 G5 Server models; it also supports the new Integrated Cluster Bus (ICB). The ICB is the core of the Parallel Sysplex Integrated Cluster, offering both improved CF interconnect bandwidth and dramatically reduced command latency to/from the CF. With the ICB technology, IBM S/390 clustered systems will be able to scale in capacity without incurring additional overhead as the individual processors increase in engine speed with each successive processor generation. The G5 Server also introduces support for the Internal Coupling channel (IC), which is a Licensed Internal Code supported "linkless" coupling channel between CF LPARs and OS/390 (TM) LPARs on the same CEC. The IC eliminates the overhead associated with LPAR-simulation of CF Coupling links previously supported via the Internal Coupling Migration Facility (ICMF). Other significant Parallel Sysplex enhancements introduced on the G5 Servers include the Enhanced Parallel Sysplex Clock functions, introducing new instructions supporting a sysplex-wide unique store clock function and support for setting different local TOD values on separate sysplexes connected to the same ETR enabling Multi-CEC Year 2000 Sysplex testing.
Internal Coupling Channels: In March 1999, the ICF or a coupling partition in G5 servers will be able to use the Internal Coupling (IC) channel to communicate between CF and ICF logical partitions and an OS/390 logical partition within a system. Model G5s using an IC channel instead of a HiPerLink channel, installations may improve overall performance of workloads in a Parallel Sysplex by up to 5%. The IC function is implemented in Licensed Internal Code (LIC) to improves reliability while reducing configuration complexity. Each IC channel is identified by 2 CHPIDs, one for sending and one for receiving. ICs have excellent performance, of up to 6 GB/sec, and can run both test and production workloads. The IC is not limited to ICFs, and can also reduce the dependency on the Integrated Coupling Migration Facility (ICMF). Up to 32 coupling connections can be supported by G5 models. ICs can use heretofore hidden CHPIDs (physically blocked), a significant value and benefit to customers nearing the 256 CHPID limit.
Finally, the IC has significant value beyond its performance characteristics; LPARs using ICs to communicate internally within a CEC (either CR LPARs or OS/390 partitions) can simultaneously use External Coupling Links or ICBs to communicate with CFs or OS/390 systems external to the CEC. This flexibility eliminates the need to "wrap" a coupling link to the same CEC to communicate internally, if external communication from the same partition is also required. This capability removes the restrictions associated with ICMF, and improves internal communication performance to the CFCC LPAR, or ICF.
Inter System Channel (HiPerLink) Enhancements: Single mode (SM) HiPerLinks have now been qualified to be used up to 10 km, (equivalent to FICON). An RPQ is available for distances up to 20 km. HiperLinks can pass data up to 100 MB/sec.
Enhanced Parallel Sysplex Clock Function:
Effective March 31, 1999, the Enhanced Parallel Sysplex Clock function will be available, including:
Note: 64 bit TOD continues to be supported (existing applications will not be impacted by this improvement).
Integrated Cluster Bus (ICB)
The ICB is the core of the Parallel Sysplex Integrated Cluster, offering both improved CF interconnect bandwidth and dramatically reduced command latency to/from the CF. With the ICB technology, IBM S/390 clustered systems will be able to scale in capacity without incurring additional overhead as the individual processors increase in engine speed with each successive processor generation. This feat is accomplished by implementing the coupling link protocols directly in hardware and exploiting an Self Timed Interconnect (STI) to STI memory bus interconnect between G5 CECs in the Integrated Cluster.
Next to the data rate of the linkless IC is the ICB. The ICB uses G5's 333 MB/sec STI to perform the S/390 coupling communication between G5 Servers. Highly integrated with very few parts the ICB is extremely reliable. Compared to a HiPerLink the ICB is a higher performing link. The 10 meter ICB cable can be used between G5 systems up to approximately 7 meters. By enabling data packets to efficiently be identified, processed and stored directly into memory the ICB is able to effectively utilize most of the high bandwidth of the STI bus (effectively 280 MB/sec of the 333 MB/sec bandwidth). The data rate improves performance by reducing the latency of synchronous send messages. This reduction in latency reduces the time a processor is idle and improves Parallel Sysplex performance, resulting in improved workload throughput. When compared to HiPerLinks ICB's, may reduce the overhead associated with coupling Systems in a Parallel Sysplex up to 3%. For example, a Parallel Sysplex incorporating a G5, with the majority of applications participating in data sharing with HiPerLinks running at 90% efficiency may be improved to 92% by use of ICB's, compared to HiPerLinks on G5. ICBs are bi-directional. ICBs will significantly reduce the cost of entry and improve the scaling of capacity in Parallel Sysplexes.
Model R06 has up to 24 STIs to connect ICBs, with a resultant reduction in the available number of channels, as ICB usage increases. The following tables identify the use of ICB's, the number of CHPID's that are reserved by their use, the remaining STI's that can be used to connect to I/O cages, and the ESCON channel maximum.
Maximum Configurations with ICB Connections: Models RA6-RD6: Models RA6, R16, RB6, R26, RC6 and RD6 have 12 STIs. Six STIs are reserved for channels, allowing these models a maximum of six ICBs.
Maximum Configurations with ICB Connections: Models R36-YX6: Models R36, R46, R56, R66, R76, R86, R96, RX6 and YX6 have 24 STIs. Six STIs are reserved for channels, allowing these models a maximum of 18 ICBs.
The following table identifies the relationship between ICBs, STIs and channels.
| # of |# of ICB|# of STI's |ESCON max. | NOTES | | ICB's|Reserved|for other |# of Chan. | | | |CHPIDs |Attachment |/Domains | | |------+--------+-----------+------------+-------| | 0 | 0 | 24 | 256 |(1) (2)| | 1 | 4 | 20 | 252 | (2)| | 2 | 4 | 20 | 252 | (2)| | 3 | 4 | 20 | 252 | (2)| | 4 | 4 | 20 | 252 | (2)| | 5 | 8 | 16 | 248 | (2)| | 6 | 8 | 16 | 248 |(4) (2)| | 7 | 8 | 16 | 248 | (3)| | 8 | 8 | 16 | 248 | (3)| | 9 | 12 | 12 | 244 | (3)| | 10 | 12 | 12 | 244 | (3)| | 11 | 12 | 12 | 244 | (3)| | 12 | 12 | 12 | 244 | (3)| | 13 | 16 | 8 | 240 | (3)| | 14 | 16 | 8 | 240 | (3)| | 15 | 16 | 8 | 240 | (3)| | 16 | 16 | 8 | 240 | (3)| | 17 | 20 | 4 | 216/8 Dom. | (3)| | 18 | 20 | 4 | 192/8 Dom. |(5) (3)| | 19 | 20 | 4 | 160/6 Dom. | (6)| | 20 | 20 | 4 | 128/4 Dom. | (6)| | 21 | 24 | 0 | 96/4 Dom. | (6)| | 22 | 24 | 0 | 64/2 Dom. | (6)| | 23 | 24 | 0 | 32/1 Dom. | (6)| | 24 | 24 | 0 | 0 | (6)|Notes:
(1) Minimum ICB configuration (2) This configuration applies to Models RA6 through YX6 (3) This configuration applies to Models R36 through YX6 (4) Maximum configuration applies to Models RA6 through RD6 (5) Maximum applicable for Models R36 through YX6 (6) Model R06 with 5, 6, 7, 8, 9, 10 ICFs only, or special plugging rules are required.
4 ESCON channels may be reduced by 1 OSA-2 or 4 Parallel channels. Max number of Parallel = 96, Parallel/ESCON/FICON = 256. ESCON channel quantities are also reduced by HiPerLinks, ICs, and ICBs.
1 ISC Adapter card blocks 4 CHPID's; each HiPerLink card uses 1 CHPID: (4 ICB's use as many CHPID's as 4 coupling links.
An IC uses 2 CHPID's -- one for sending and one for receiving
To summarize this section, all Generation 5 models can have up to 32 coupling connections. Models RA6-RD6 can have up to a maximum of 6 ICB's. Models R36-YX6 can have up to 18 ICB's. Model R06, a dedicated coupling model with no ESCON, FICON, Parallel, or OSA-2 channels, can have up to 24 ICB's. A total of 32 links in combination of HiPerLinks, ICB's and IC's is possible on any G5.
Coupling Facility -- Options
An IBM Coupling Facility (CF) is always defined as a PR/SM LPAR partition. For a 9674 or a 9672-R06, only coupling-facility partitions may be defined. For all other 9672 machine types, both OS/390 and CF (that is, sender and receiver) partitions may be defined on the same machine. Sender and receiver partitions may additionally be defined as enabled for the Integration Coupling Migration Facility (ICMF).
Depending on the 9672 model, the CF processors defined to a particular CF partition may be dedicated internal-coupling-facility (ICF) processors, dedicated central processors (CPs), shared CPs, or dedicated ICF processors and shared CPs (dynamic ICF expansion). When CPs are defined, the customer is subject to software charges for all MSUs or the models software group, including the engines which may be doing the coupling activity. ICF processors are not subject to this charge.
There are four types of coupling links available, depending on the machine type and model: ISC, HiPerLink, ICBs, and ICMF links. The choice of links is dependent on configuration, distance, and performance requirements. ICMF links may only be used when the sender and receiver partitions are both enabled for ICMF and they are on the same physical machine. In this configuration, no other coupling links are available for these partitions.
Internal Coupling Facility (ICF) Value, (Feature #0991)
The ICF, which utilizes a spare Processor Unit (PU), can be a Coupling Facility option. ICF, introduced on G3, extended on G4 in quantity and function, grows in value with its implementation on G5. Most G5 models will be able to configure ICF's on them. Only on G5 models that do not have any extra spare PUs is the ICF feature not offered (models RD6, R96, RX6, YX6 have 0 ICF's). In April of 1998 the maximum number of ICFs on G3/G4 was increased to 4 (based on number of PU's not used). G5 can have more ICFs than either G3 or G4. Up to 7 ICFs in the R36 and up to 10 on the R06 can be configured. Effective March 1999, the maximum number of ICFs will be increased to 8; models RD6 and R96 will have the option of being configured with 1 ICF (no spare PU).
As an option to reduce Coupling Facility cost of ownership, the ICF has both characteristics of a 9672 CF and a stand alone CF. Because the ICF is a CF partition configured to run on only the spare processor(s) of a 9672, it shares the exclusion of software license charges like the R06 or a 9674. Coupling capacity that does not incur software charges lowers the cost of configuring systems in a Parallel Sysplex. By using the Server's infrastructure (power, cooling, mechanical, memory, logic, service element) for coupling, an ICF reduces the power, cooling and space costs versus what a standalone Coupling Facility such as a 9674 or a R06 would be. An ICF also has reduced maintenance charges over a standalone Coupling Facility which has thousands of parts to maintain in addition to the Coupling Facility Control Code. The reduced maintenance price of an ICF reflects the lower maintenance required by an ICF, over a Stand Alone Coupling Facility (9674 or R06).
When using an ICF as a Coupling Facility, it is advantageous to have 2 or more OS/390 partitions to take advantage of continuous operations protection from software outages. Individual OS/390 partitions can be taken down for maintenance or OS/390 release upgrade, without suffering application outage. This is done through data sharing provided by the remaining LPARS in the System.
Dynamic ICF expansion allows an ICF logical partition to acquire additional processing power from a pool of shared CP's executing normal OS/390 production or test work for the system. This has been called an L shaped LPAR for CF's. This capability makes the ICF an ideal backup for either a standalone coupling facility or another Systems ICF(s) which may have more coupling capacity. If an ICF processor fails, even running dedicated, a spare PU can be dynamically switched in to substitute for the failed processor, transparent to the CFCC Licensed Internal Code running on that processor.
G5's ICF is attractive for both test and production workloads. G5 has improved coupling connections. One in particular, the Internal Coupling (IC) channel is ideal for an ICF's coupling to the OS/390 partition(s) inside the same CEC with more efficiency than any other coupling link. The linkless IC couples with more efficiency and in a more cost-effective manner than physical channels because there is no set up for LPAR code or coupling links. The IC reduces the cost of Parallel Sysplex while improving its performance. G5's other new coupling connection, the Integrated Cluster Bus (ICB), has approximately 280 MB/sec bandwidth or 3 times the maximum bandwidth of the improved coupling links on G3 and G4, HiPerLinks. ICB's connecting G5's and the use of ICF's reduce the cost and complexity of Parallel Sysplexes while improving performance.
ICF's are ideal for coupling Resource Sharing, or Systems Enabled Parallel Sysplexes; sysplexes that are not in production data sharing with IMS (TM), DB2 (R) or VSAM/RLS. Simplified systems management can be achieved by using XCF structures versus ESCON Channel to Channel connections. Improved scalability is possible while Systems Enabled when either GRS STAR or RACF (R) is used. With GRS Star, the traditional ring mode protocol for enqueue propagation is replaced by a star topology where the CF (or ICF) is the hub. By using the ICF and GRS STAR, enqueue service times can be reduced up to 10X. RACF can leverage command propagation in a Sysplex to refresh each System's copy of the RACF database. One step further is when RACF caches its entire database in the CF. This provides high speed access to security profiles.
ICF's such as either a standalone R06 or a 9674 Coupling Facility need configuration planning must still account for memory and links. ICF's will likely increase memory requirements, especially if software exploits the CF to provide additional function not available except when running a CF in a Parallel Sysplex.
A production Systems Enabled environment with 2 ICF's will not compromise sysplex availability or integrity. The following is an example or 2 systems each with an ICF:
| System 1 System 2 | |----------------. .--------------| |OS/390 Image 1 | Coupling |OS/390 Image 2| |----------------+-------------+--------------| |ICF | Connection |ICF | |XCF Signaling | (ICB or |XCF Signaling | | Logger Operlog | HiPerLink) |Logger Logrec | | RACF Primary | |Shared Tape | | Batch Pipes | |RACF Secondary| | | |GRS | | JES2 | | |
The combination of an ICF and the following software in a Systems Enabled (Resource Sharing) Parallel Sysplex produces significant benefits to customers without full datasharing as noted in the following table:
Benefit | Function |Software |
| |Exploiter |
------------------+-----------------+--------------|
Load Balancing | Cross-system | SmartBatch |
| BatchPipes (R) | |
------------------+-----------------+--------------|
Resource Sharing | Shared Tape | OS/390 |
| | Allocation |
------------------+-----------------+--------------|
Performance |High Speed | RACF |
|access to | (Security |
|security Profiles|Server) |
------------------+-----------------+--------------|
Improved ENQ/DEQ |Resource | OS/390 GRS |
performance |Serialization | |
------------------+-----------------+--------------|
Systems | | |
Management |Checkpoint | JES2 |
------------------+-----------------+--------------|
Improved |OPERLOG and | OS/390 Sys. |
Systems Mgt |LOGREC logstream | Logger |
------------------+-----------------+--------------|
Simplified System |High Speed | OS/390 XCF |
Definition |Signalling | |
------------------+-----------------+--------------|
Availability |MNPS | VTAM (R) |
|(needs APAR) | (multi node |
| | persist.scan)|
------------------+-----------------+--------------'
For additional information on exploiters see CF Configuration Alternatives: A Positioning Paper (May 1998).
For the present, 2 ICF's are the preferred solution for Resource Sharing Parallel Sysplexes and either 2 Standalone CF's or 1 Standalone CF and an ICF for backup are the preferred solution for full data sharing (IMS, DB2, VSAM/RLS) Parallel Sysplexes.
One of the latest innovations is the G5 capability to upgrade a coupling ICF to an Application CP. This chargeable PU upgrade has the effect of "removing the ICF": it allows a customer to buy to buy another ICF on a different system, at standard pricing, to effectively remove an ICF and install an ICF on another machine. Conversely, the G5 has the capability to convert an Application CP to an ICF via an MES, when CF consumption of the Application CP is high enough to warrant conversion to a dedicated ICF processor (this action may avoid on-going software license charges on that processor).
9672 Model R06 Value
The R06 is an option that a customer may use to define Coupling Facility partition in a standalone environment. For Application Enabled (data sharing) Parallel Sysplexes it is strongly recommended that one or two R06's be utilized. The 9672 R06 has the flexibility of operating similar to a 9674 dedicated coupling facility today and in March 1999, upgrading to another Generation 5 Server. The R06 uses the latest level of Coupling Facility Control Code, CFCC.
A standalone coupling facility assists configuring a Parallel Sysplex to be resilient to failure. This means that the loss of a Coupling Facility would not force recovery of CF resident data through the use of logs or even lose data where no logs exist. Recovery through use of logs is a lengthy process and may take hours, so a standalone CF (either 1 or 2) are needed for the Application Data Sharing environment.
All of the R06's resources are devoted to coupling. As a standalone CF it can be independently upgraded with minimal planning and not impact production workloads. The R06, not running OS/390 or customer workloads, has value not only in physical isolation that dedicate 100% of its resources to coupling, but it also saves software costs by not being charged for running applications; moreover, the R06 is consistent with the objective for CMOS Servers (with ICF's) being Cross Connected (mutually supporting) and capable of all the functions on today's standalone CF's.
The R06 has more coupling and link capacity than any other CF. R06's can scale to 10 coupling engines, more than the next highest model, the R36's 7 ICF's (increases to 8, effective March 1999). The R06 also can attach up to 24 ICB's and have up to 24 GB of memory (memory capacities in GB are: 1, 1.5, 2. 2.5, 3, 3.5, 4, 5, 6, 7, 8, 10, 12, 16, 20 and 24 GB). The types of coupling connections on R06 include ICBs and HiPerLinks.
9672 Model R06 Minimums and Maximums: The 9672 R06 model replaces 9674's as a standalone Coupling Facility. R06 hardware is dedicated to the coupling tasks. R06's are not charged for OS/390 or application software. The R06 uses the same technology as other Generation 5 models. Because there is no need for Parallel, ESCON, OSA-2 or FICON Channels, the R06 has the ability to scale Integrated Cluster Busses, the new G5 Coupling connections higher.
The following table compares and contrasts features of the R06 with the other G5 Models:
Feature/Function|R0* (1-4W)|R0*(5-10w)|RA6-RD6|R36-YX6| ----------------+----------+----------+-------+-------| ICB's: | | | | | min. | 0 | 0 | 0 | 0 | max. | 12 | 24 | 6 | 18 | ----------------+----------+----------+-------+-------| HiPerLinks: | | | | | min. | 0 | 0 | 0 | 0 | max. | 32 | 32 | 32 | 32 | ----------------+----------+----------+-------+-------| IC's: | | | | | min. | 0 | 0 | 0 | 0 | max. | 0 | 0 | 32 | 32 | ----------------+----------+----------+-------+-------| Coupling Con- | | | | | nections (IC, | | | | | ICB, HiPerLink):| | | | | min. | 1 | 1 | 0 | 0 | max.(one/ICF) | 32 | 32 | 32 | 32 | ----------------+----------+----------+-------+-------| Internal Coupl. | | | | | Facilities | | | | | (coupling PU's):| | | | | min. | 1 | 5 | 0 | 0 | max. | 4 | 10 | 3(RA6)| 7(R36)| ----------------+----------+----------+-------+-------| Channels (Paral-| | | | | lel/ESCON): | | | | | min. | 0 | 0 | 3/4 | 3/4 | max. | 0 | 0 | 256* | 256* | ----------------+----------+----------+-------+-------'Note: Although 32 pairs of ICs can be configured (for a total of 64 sending and receiving required) it is not recommended as ICs have high bandwidth and large numbers of ICs are not necessary.
(*) The 3 channel minimum only applies to MES's that bring forward the older 3 port parallel channel card or systems shipped in 1998 prior to availability of feature 2304 (4 port parallel card). All new build G5's have 4 port Parallel and ESCON cards and their minimum is 4 channels. It is important to realize that I/O options are not additive to the 256 channel connections. There is a reduction in 256 ESCON channel maximum for each Parallel channel card, OSA-2, IC, ICB, HiPerLink or FICON connection.
The use of PU's for SAPs and spares is also different for the R06. Every R06, 1way to 10way, has a spare PU; All R06's have a single SAP. Models RX6 and YX6 have no spares. Models R86, R96, RX6 and YX6 have as standard 2 SAP's to drive the larger I/O workloads associated with 8 or more processors. R06's are plugged like other 9672 G5 models to minimize CHPID disruption (changes) when upgrading from an R06 to other Generation 5 models.
Memory by model (GB)
| Model:| R06 | R06 | RA6-RD6| R36-RX6 | | | (1-4 ICF)| (5-10 ICF)| | | |--------+----------+-----------+--------+---------| |minimum | 1 | 2 | 1 | 2 | |maximum | 12 | 24 | 12 | 24 |
Memory increments for R06 and other models are identical. Upgrades to model 9672 R06 from 9674 models C04 and C05 are available September 30, 1998; upgrades of model R06 to the other G5 server models will be available March 31, 1999, introducing new flexibility to preserve asset value. Future upgrades of R06's will be tied to the model structure and option limits of the model the R06 is upgrading to. R06's have their I/O plugged like other 9672 R6's in order to minimize CHPID changes. Deviations to plugging rules require an RPQ.
Significant New S/390 Architecture Instructions
S/390 G5 Parallel Enterprise Server architecture has 121 new hardware instructions beyond prior S/390 CMOS models. Major categories of instructions include:
Java is a Trademark of Sun Microsystems Inc. (*) ANSI/IEEE Standard 754-1985, IEEE Standard for Binary Floating-Point Arithmetic.
G5 Security Advantages
Secure Consolidation with PR/SM
The Processor Resource/System Manager* (PR/SM*) facility of the G5 Server family has been submitted for evaluation against the Information Technology Security Evaluation Criteria (ITSEC) at an E4 assurance level. The evaluation is being conducted in Germany by the IABG CLEF.
The E4 level provides a very high degree of assurance that PR/SM can be used in environments where separation of workloads is a requirement, but where the use of a single hardware platform is desirable for reasons of economy, flexibility, security, or management. PR/SM provides for secure isolation by preventing the flow of information among partitions. This isolation may be used where the separation is based on need to know, or where data at differing security classifications must be isolated.
This certification validates the G5 processor as an essential building block for business solutions integrating e-Business applications and traditional corporate workloads on a single platform.
The certification is expected to be completed in fourth quarter 1998.
Cryptographic Coprocessor: IBM's leading edge Cryptographic Coprocessor is the world's fastest dedicated cryptographic engine for scientific and commercial workloads. The highest level of security for protecting information , Triple DES, is now available worldwide using IBM's S/390 CMOS Cryptographic Processor.
The S/390 CMOS Cryptographic Coprocessor is exportable and is enabled by a unique diskette that is linked to the systems's individual serial number. Cryptographic coprocessors are standard on every G5. Every G5 model is built with 2 cryptographic coprocessors. Even model R06 has 2 Cryptographic coprocessor chips which can be enabled if the R06 upgrades to other models. G5's crypto chips are tamper proof. G5's crypto engines support a variety of cryptographic standards including 40, 56, 112 and 168 bit algorithms for protecting information.
Public keys standards RSA and DSS (DSS is used in many U.S. Government bid specifications) are supported.
The following new functions announced on February 24, 1998, for the Generation 4 are also available on Generation 5 to enhance IBM's commitment to Network Computing Framework (NCF) including the security for IBM's Net.commerce initiatives, emphasizing e.commerce.
These new functions available with OS/390 V2R5 are:
Secure Electronic Transaction (SET) is a protocol developed jointly by VISA International and MasterCard for safeguarding payment card purchases made over open networks. It has become pervasive in use by e-commerce as a transaction protocol and software vendors are incorporating SET into existing browsers and merchant servers.
ICSF has been enhanced to include support for SET by implementing RSA-OAEP encryption. This support will enable vendor and user-written applications to take advantage of the integrated crypto capabilities of S/390. This includes:
VISA and MasterCard have specified a cryptographic method to calculate a value that relates to information on a payment card such as the personal account number (PAN), the card expiration date, and the service code. The use of card-verification values (CVV) can reduce the risk of losses resulting from alteration/counterfeiting of the card's magnetic stripe.
ICSF has been enhanced to provide callable services to generate and verify the VISA card-verification value (CVV) as well as the MasterCard card-verification code (CVC) using the track-2 method. This support provides additional card integrity for financial processes involving payment cards. CVV/CVC functions are available on DES (non-CDMF) systems only.
ICSF now provides a storage data set for Public Key Algorithm (PKA) private and public keys. Storage of PKA keys within a Public Key Data Set (PKDS) allows existing PKA APIs to specify a record label as an alternative to specifying a key token and eliminates the need for PKA applications to provide their own PKA key storage mechanism. Customers will now be provided with the same security, flexibility and efficiency working with public/private keys as they have had with DES keys.
Padding using zeroes as an alternative to padding according to Public Key Cryptography Standard (PKCS 1.2) is now available when using the Symmetric Key Generate, Symmetric Key Import, and Symmetric Key Export functions. The services will each allow for and process a new keyword, ZERO-PAD, in the rule array.
The ANSI X9.19 Optional Double-MAC Procedure is supported by ICSF. This requires the use of a double-length key for MAC processing. This support addresses customer requirements for enhanced security for message authentication. Emerging VISA International and MasterCard protocols in Europe also call for ANSI X9.19 MAC. Generation of both user derived keys and SESSION MAC keys will be supported.
DES encryption for ICSF has been enhanced to now include TDES (168 bits) encryption. This availability of strong encryption provide customers with increased flexibility in choosing the level of data encryption they desire. Finally, the TDES performance of the G5 has been significantly improved over the performance observed on G4 models.
Exploitation
Application and Product Utilization of Hardware Encryption: There are a number of different applications, products and communication protocols which rely on encryption to offer data confidentiality for protecting sensitive information and data integrity for ensuring information accuracy. In addition, many international standards and defacto standards have architected secure communication options based on encryption such as:
Migration/Coexistence Considerations: When migrating from ICSF V1.1 to ICSF V2.1, one must first migrate the CKDS to the current format. A migration utility called CSFCVR1 is provided. The utility can be run on either ICSF V1.1 or on the latest release.
When migrating from ICSF V1.2 to ICSF V2.1, no migration is necessary but new system keys must be defined for CKDS if new key functions are to be used. Once new system keys are added to CKDS, CKDS is not sharable with ICSF systems which do not support those key types.
ICSF V2.1 PKA implementation differs in some respects from the implementation in the Transaction Security System (475x) products.
In the S/390 G5 Server, TDES support is in LIC Driver 10 and beyond. The S/390 G5 Server joins the G4 as the only two processors supporting TDES for privacy.
Requirements: These enhancements require OS/390 V2 R5 and beyond.
Export Considerations: As with the previous announcement in June 1996, regarding the CMOS Cryptographic Coprocessor, an enablement diskette will be shipped with the server to customers who decide to enable the Crypto Coprocessor.
Customers located within the USA and Canada can receive TDES functionality.
Banking and Finance customers located outside the U.S. and Canada can receive TDES under a blanket licensing agreement that IBM has with the U.S. Dept. of Commerce. All other customers outside the USA and Canada must have an export license in order to be given this level of encryption. In all cases, contact your IBM Export Regulations Office for specific guidance.
Availability Enhancements
S/390 Open Systems Adapter 2
One S/390 Open Systems Adapter 2 (OSA-2) is shipped with each S/390 G5 Parallel Enterprise Server providing seamless, scalable connectivity to Local Area Networks (LANs). The OSA-2 feature plugs into an I/O slot and has the same reliability and availability characteristics of a S/390 channel. Any of the media types can be selected up to a maximum of 12 features for each G5 Server. The selectable features are as follows:
Refer to Hardware Announcement 198-040 dated February 24, 1998, (Fast Ethernet Plus New Functions for S/390 Open Systems Adapter 2; S/390 Parallel Enterprise Servers G3 and G4 Enhancements.
For additional information on all of the S/390 Open Systems Adapter 2 features and the S/390 Open Systems Adapter Support Facility, contact your IBM representative.
Year 2000
These products do not have date dependencies and are therefore Year 2000 ready.
The maintenance end date for this year 2000 ready product is September 30, 2006.
IBM's Large Server Classic Strengths:
Why S/390? S/390 Parallel Enterprise Servers have enabled a computing environment with the best of breed electronics and information technology, to offer IBM's customers:
Progress Report: The Transformation of S/390
IBM kept what was right (reliability, scale, availability, security, & manageability), and then shrank it with CMOS, then opened it (UNIX, ATM, CORBA, TCP/IP, IEEE, DCE, Domino, NT, Java, SAP R/3, adapted it Peoplesoft, and so forth), networked it (Domino Go Webserver, NETWORK STATION (TM), and so forth) and clustered it (Parallel Sysplex). We made it self-tuing (Workload Manager) and integrated everything (OS/390) so that it supports the industry growth initiatives (e-business, Server Consolidation, Business Intelligence and Applications. The following sections detail progress:
As a clarification of IBM's October 7, 1997 Medium Sized Enterprise Announced intent to make applications easier to run in a S/390 environment, we are providing the following additional information. Secure Network Computing and e-business, Lotus (R) Domino for S/390, the hottest UNIX based Enterprise Resource Planning (ERP) applications, Server Consolidation of UNIX application workloads, and the ability to run generalized UNIX interactive workloads will now be within easy reach of S/390's Medium Sized Enterprise Customers. This can provide a lower total cost of computing solution for small or departmental solutions in large enterprises.
For Medium Sized Enterprise Customers, these exciting new capabilities will be enabled and made available via an easy-to-use OS/390 automated UNIX application server which will be available as an additional OS/390 delivery vehicle option. For VM and VSE customers, new offerings similar to the current IBM OS/390 Implementation for VM and VSE Offering will help make this new capability financially attractive. The automated UNIX application server will coexist on the same processor in either an LPAR or as a VM Guest machine with the current VM and/or VSE systems. Existing VM and VSE customer applications can continue to run unchanged. Traditional VM/VSE system programmer and operator skills are required, but, the easy-to-use OS/390 automated UNIX application server's objective is to require no OS/390 (MVS) systems programmer or operator skills to install, maintain or use. UNIX skills for the easy-to-use systems administration interface of the automated UNIX application server, UNIX based ERP applications, and UNIX workloads will be sufficient to manage the applications.
With the Generation 5 CMOS Servers, IBM has improved capabilities of existing Parallel Sysplexes involved in full data sharing of IMS, DB2 and VSAM/RLS workloads as well as migration to Parallel Sysplex.
Generation 5 coupling engines are larger and have improved price/ performance. R06s and ICF allow granular capacity additions. The new IC and ICB connections, only on G5 models, gives a lower cost of entry to enable Parallel Sysplex configurations in establishments.
The ICB is the core of the Parallel Sysplex Integrated Cluster, offering both improved CF interconnect bandwidth and dramatically reduced command latency to/from the CF. With the ICB technology, IBM S/390 clustered systems will be able to scale in capacity without incurring additional overhead as the individual processors increase in engine speed with each successive processor generation. Because this combination of ICF/IC performance may be orders of magnitude higher than ICMF, it is recommended that customers use ICF/IC as their entry to Parallel Sysplex configurations, for a smooth transition from test to production.
Clarified Criteria for Aggregated PSLC pricing
On January 13, 1998, IBM announced clarified criteria to qualify for aggregated Parallel Sysplex License Charge (PSLC) software pricing. IBM has always required processors to be "actively coupled" to qualify for aggregated PSLC software prices. This announcement adds qualifying criteria to IBM software customer agreements that clarifies what it means to be "actively coupled".
IBMers and customers have asked for more details of what must be done to qualify for aggregated PSLC prices. The lack of detailed criteria has led to inconsistencies in determining when customers qualify for aggregated PSLC pricing. The objective of this announcement is to clarify the criteria so it can be consistently applied to all customers.
In addition to having qualifying hardware and software, all processors in a sysplex must be physically attached to a common Coupling Facility and timer. All of the MVS or OS/390 images in the sysplex must have at least one common systems enablement function, of the eleven functions that qualify, activated to use the Coupling Facility across all images in the Sysplex. In addition, this must be the normal mode of operation for the sysplex and the OS/390 and MVS images participating in the above functions must account for at least 50% of the total OS/390 and MVS total workload on each machine.
Customers currently receiving aggregated PSLC pricing will be given a grace period up to the end of September, 1998 to meet the clarified criteria. Grace periods may be shorter if the customer's implementation is simple. Customers adding new sysplexes must meet the criteria now to qualify. Exceptions to the criteria will be considered for customers committed to a parallel implementation strategy that have valid and compelling reasons for not meeting the criteria. Exceptions will be handled through the software special bids process.
Scalable price performing coupling engines and connections are part of IBM's commitment to enabling customers to enjoy the benefits of Parallel Sysplex. SOFTWARE LICENSE CHARGES HAVE BEEN MODIFIED TO REFLECT PARALLEL SYSPLEX USEAGE.
S/390 has industry-leading levels of availability, security capacity and reliability for enterprise computing needs. Many of you have heard and know about IBM's e-business strategy. To conduct secure e-business you need:
BI workloads are resource intensive and at the same time very unpredictable. Short running queries coexist with extremely demanding complex analyses, data mining runs and updates to the warehouse. OS/390 Workload Manager (WLM) is able to manage this diverse workload using business priorities so that the customer service rep's short running query does not wait behind the fraud detection data mining run. WLM is able to deliver guaranteed response time to more important work and distribute the remaining resources among the other work, based upon desired goals and importance. that can terminate important work midstream just because it has exceeded an arbitrary threshold. WLM can be set to automatically downgrade the importance of the work as it ages in the system and prevent it from degrading other queries. Functions like these mentioned enables OS/390 to support many more users than other BI platforms, making its cost /user lower than many UNIX platforms (based on ITG cost of scalability).
Customers like Edison International are using OS/390 and DB2 to provide themselves with a competitive advantage in the increasingly deregulated utility industry. Edison has been building a warehouse and data marts on OS/390 for a number of years. They have data from many parts of their business (human resources, power usage, customer information, and so forth) in their DB2 BI system. A financial study of one of their data marts (a credit collections data mart) by IDC revealed a return on investment of 662%. Edison makes the BI data available to all their employees using a large variety of tools, both traditional and new (and so forth, SAS) MS Access. They also have web-enabled their warehouse, making it accessible via the intranet. Edison picked S/390 because it enabled them to scale easily while utilizing their existing investment in skills and tools.
S/390 Applications Initiative
The IBM S/390 Applications Initiative is directed toward providing support for customers in three key areas:
Enhanced Support for Existing Applications
Support is provided by both hardware and software improvements:
The C/C++ compiler on OS/390 continues to utilize the latest levels of IBM optimization technology for each of its new releases and improvements have been made to DLL processing. Enhancements in the Kernel and Run Time Library (RTL) have significantly improved the integration of Unix services within the operating system and reduced the overhead to invoke these services by 80% between OS/390 R3 and R4 and by an order of magnitude from the early POSIX releases. Web serving throughput has increased by 100 fold since the first release of the Internet Connection Secure Server (now called Domino Go WebServer) and continued large scale performance enhancements are planned for 1998 and 1999.
New Software Technology to Enable New Applications
The performance of open connectivity to DB2 is greatly improved with DB2 5.1 through the implementation of Store Procedure Calls under Work Load Management (WLM), native TCP/IP sockets, improved dynamic SQL processing, and the avoidance of ASCII/EBCDIC translation. In addition, profiling tools are being created to assist in performance optimization of C/C++ applications.
With the above combination of pathlength scrubbing, reduced layering, component restructures, and new hardware instructions, we anticipate large improvements in overall end-to-end S/390 UNIX performance in 1998 and 1999.
Applications New to the S/390 Platform
A variety of new applications have become available on S/390 and OS/390 in recent months. Key among these is SAP R/3, which is one of the leading Enterprise Resource Planning (ERP) applications. These enhancements continue to support UNIX applications on OS/390. In particular, the increased capacity of the new G5 processor and its associated I/O subsystems will provide improved performance for SAP R/3.
In addition, the database implementation of SAP R/3 on S/390 transparently inherits the datasharing strengths of the parallel sysplex environment. This means that customers can achieve open ended scaling capabilities in the S/390 SAP database environment. In addition to scaling benefits, the availability characteristics of DB2 in a parallel sysplex environment are also intrinsically part of the SAP R/3 implementation on S/390. From a flexibility perspective, there are connectivity options of using either OSA or ESCON, and to link in either NT or AIX application servers. Finally, for existing S/390 customers, the ability to leverage investments in the platform and skills when implementing an SAP solution, while gaining the availability and scalability attributes previously described make an extremely attractive solution. The IBM S/390 Parallel Enterprise Server (TM) is the industry leader for running multiple diverse, mission-critical workloads including online transaction and batch processing, financial and administrative applications, business intelligence and new e-business applications. Industry consultants estimate that more than two thirds of the world's business data resides on IBM Servers, most of it is on S/390.
S/390 has industry-leading levels of availability, security capacity and reliability for enterprise computing needs. Many of you have heard and know about IBM's e-business strategy. To conduct secure e-business you need:
With OS/390 Version 2, IBM will follow through on that announcement and deliver Component Broker on the S/390 platform. Initially, this will be done for a limited set of customers participating in concept validation on OS/390 Version 2 Release 5. At the conclusion of concept validation, Component Broker for OS/390 will be orderable no-charge for those customers with OS/390 Version 2 Release 5.
Component Broker for OS/390 will provide an environment for the hosting of new business applications on S/390 that, using a new programming model, will enable customers to rapidly design and develop their applications. This programming model will be the same across platforms.
Component Broker for OS/390 will leverage S/390 strengths such as Parallel Sysplex technology for scale and availability, and will exploit the OS/390 Workload Manager for proper workload balancing and management. This will enable Component Broker work to run along with traditional workloads such as IMS and CICS.
The G5 server with its increased I/O bandwidth and the FICON strategy will provide added robustness to the overall object-oriented environment on S/390.
Additional information concerning Component Broker for OS/390, can be found on the World Wide Web at the OS/390 Home Page and the Component Broker Home page at the following URL:
Parallel Sysplex -- Introducing the GEOPLEX
The GEOgraphically dispersed parallel sysPLEX (GEOPLEX) is a parallel sysplex spread across 2 or more sites up to 40 km apart; all critical data is remote copied between the sites. The GEOPLEX provides the capability to manage the remote copy configuration, and storage subsystem, in addition to being able to automate planned reconfigurations or failure recovery from a single point of control. The GEOPLEX supports all TP monitoring (such as, CICS & IMS, and all DB managers (for example, DB2, IMS, & VSAM). In the event of a planned system or site shutdown, GEOPLEX will quiesce affected workload (and optionally restart it elsewhere), while the workload continues to execute on other nodes. In the event of a site or system failure (for example, a disaster), GEOPLEX will restart the failed workload on surviving nodes, while work continues to execute.
IBM has proven this concept with multiple Joint Customer Studies including IT Austria, the newly formed IT Subsidiary of Bank Austria, Creditanstalt, Erstebank, and Spardat in Vienna, Austria, establishing a dispersed, multi-site environment for application programs using facilities enabled with:
A White Paper, "GEOPLEX, a Multi-Site Availability Environment for Application Programs" that details IBM's approach to satisfy continuous availability solutions, is available from your IBM Marketing Representative, or can be downloaded from the Internet URL:
STATEMENT OF GENERAL DIRECTION
Notice
Actual performance and environmental costs will vary depending upon individual customer configurations and conditions. All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.
Trademarks
PR/SM, OS/390, IMS, MVS, CommercePOINT, MVS/ESA, NETWORK
STATION, Multiprise, Intelligent Miner, S/390 Parallel
Enterprise Server, and eNetwork are trademarks of International
Business Machines Corporation in the United States or other
countries or both.
S/390, Parallel Sysplex, ES/9000, ESCON, DB2, RACF, BatchPipes,
VTAM, CICS, ThinkPad, APPN, Nways, AIX, System/390, and RAMAC
are registered trademarks of International Business Machines
Corporation in the United States or other countries or both.
Java is a trademark of Sun Microsystems, Inc.
UNIX is a registered trademark in the United States and other
countries exclusively through X/Open Company Limited.
Domino is a trademark of Lotus Development Corporation.
Lotus is a registered trademark of Lotus Development
Corporation.
Other company, product, and service names may be trademarks or
service marks of others.
Information about the S/390 (R) G5 Parallel Enterprise Server processor models can be found in the following publications which are now available:
Order
Title Number
System Overview GA22-7158
S/390 G5 Enterprise Server GC22-7106
Installation Manual --
Physical Planning
The following publications are available to be ordered prior to shipment of a G5 model:
Order
Title Number
Pre-Installation Configuration GC38-3120
Workbook (G5 models)
Learning to Use the S/390 CMOS SK2T-2512
Console
Introduction to IBM S/390 FICON (TM) SG24-5176
IBM S/390 FICON Migration Guide SG24-5169
The following publications reflect the announced product and are shipped with the product:
Order
Title Number
PR/SM (TM) Planning Guide GA22-7236
IOCP User's Guide GC38-0401
Managing Your Processors GC38-0460
Hardware Management Console Guide GC38-0470
Operations Guide GC38-3119
Standalone IOCP User's Guide GC38-0458
Pre-Installation Configuration GC38-3120
Workbook (G5 Models)
HWMCA Programming Interfaces SC28-8143
Problem Analysis (PA) Guide SY22-9876
Specified Operating Environment
Physical Specifications: 9672 Servers
The mechanical package for the S/390 9672 Parallel Enterprise Servers -- Generation 5 conforms to EIA guidelines for frames. Packaging for new build systems may be in one or two frames, depending on the configuration of the system. Packaging for model upgrades is in one or two frames.
Environmental Specifications -- Server Models
New Build New Build
Minimum Maximum
1 Frame 2 Frame
System System
Power Requirements
50/60 Hz, KVA 0.6 5.5
Heat Output KBTU/hr. 2.0 18.8
Air Flow CFM 290 1400
Air Flow m*3/min 7.1 38.6
Floor Space
-- Sq meters 1.0 1.8
-- Sq feet 10.4 19.7
Including service clearance
-- Sq meters 2.5 4.8
-- Sq feet 27.4 51.9
Approximate weight
-- Kg 612 938
-- Lb 1346 2057
9672 R06 Coupling Facility
The mechanical package for the S/390 Coupling Facility is designed in conformance with EIA guidelines for frames. The 9672 R06 model is packaged in one frame.
New Build New Build
Minimum Maximum
Power Requirements
50/60 Hz, KVA 0.6 2.6
Heat Output KBTU/hr. 2.1 8.9
Floor Space
-- Sq meters 1.0 1.0
-- Sq feet 10.4 10.4
Including service clearance
-- Sq meters 2.5 2.5
-- Sq feet 27.4 27.4
Approximate weight
-- Kg 612 620
-- Lb 1346 1364
Standard Hardware Functions and Capabilities
Supported I/O Devices
The I/O devices and control units supported for attachment to the IBM OEMI/parallel and ESCON channel interface on the 9672 G5 models are the same as currently supported on the 9672 R1, R2, R3, G3, and G4 models. For a complete list of the supported I/O for the G5 models, refer to the 9672 Sales Manuals.
Hardware Management Console (HMC)
The Hardware Management Console provides a single point of control and single system image for managing local or remote hardware elements for the 9672 systems. The Hardware Management Console(s) communicate with the 9672 system(s) using either an SNA protocol or TCP/IP protocol flowing over either a Token-Ring LAN or over an Ethernet LAN. The Hardware Management Console employs a state-of-the-art direct-manipulation object-oriented graphical user interface supporting exception based real-time system status reporting via ICONs and colors, consolidated hardware messages, consolidated operating system messages, and consolidated service support (including modems and phone lines), and of course, full operation of the managed systems.
The Hardware Management Consoles ordered with a new S/390 G5 Server will be shipped as feature code #0041. All MES orders will include either replacement of the existing Hardware Management Console by #0041 or an upgrade to #0031, depending upon the level of system in the existing Hardware Management Console. HONE AID (CFSYSTEM) will determine the appropriate shipment content. When upgrading a system which has #0009 installed, that Hardware Management Console will be upgraded to either a #0031 or #0041, depending upon the type of PC in the existing Hardware Management Console (PS/2 (R) based #0009 will be exchanged for #0041; Model 6885 based #0009 will be upgraded to #0031). All upgrades from 9672 G3, 9674 C04 or earlier models, presently using #0021 should confirm order of FC #0031, as a no-charge feature. All upgrades from 9672 R3 as well as 9674 C03 models presently using #0009 should confirm feature exchange for either #0031 or #0041, depending upon the type of PC in the existing Hardware Management Console, as a no-charge feature exchange. #0041 is also available as an optional feature for customers who desire additional Hardware Management Console performance and flexibility, especially in multi-system environments. No upgrade path from #0031 to #0041 is available; #0031 and #0041 are fully compatible and may coexist on the same LAN. #0023 only be ordered if the #0041 Hardware Management Console will be connected to a Token-Ring LAN.
Technical differences between #0031 and #0041 are as follows:
Support Element: The G5 server models contain an integrated Support Element (SE) which provides distributed hardware operations and management. The SE controls and monitors the operation of the 9672 and sends status, hardware messages, and operating system (console integration) messages to the Hardware Management Console(s) for consolidation and exception processing.
Remote Operations: The System/390 (R) Parallel Enterprise Server and Coupling Facility hardware systems management products support remote operations in a variety of ways over a variety of communications connections. In each case, the objective is to enable a human or programmed operator to monitor or control a remote system in essentially the same manner as if the operator were at the same site as the remote system. Remote operations include:
Several facilities and Application Programming Interfaces (APIs) are available to other management applications to allow for remote operations and monitoring. These APIs support TCP/IP and SNA based remote management applications which may be provided by IBM, other software vendors, or the enterprise.
The Hardware Management Console provides a system management set of APIs by means of the TCP/IP SNMP protocol.
The Support Element provides a system management set of APIs by means of the TCP/IP SNMP protocol.
The Support Element provides a SNA based interface supporting SNA based management applications such as IBM NetView (R) and IBM TSCF.
HWMCA Programming Interfaces: In addition to providing an end-user with the ability to view and manipulate managed objects, the Hardware Management Console also provides APIs. The APIs are:
Customization: The HMC can be tailored to the specific needs of each individual user. The variety of ways that the HMC can be customized include:
Access to the OS/2 desktop and other applications residing on the HMC can be limited for individual users.
Control of which managed objects, such as hardware elements, operating system images, and groups, that are accessible for individual users.
Control of which HMC tasks that are accessible for individual users.
OPTIONAL FEATURE DESCRIPTIONS
Server Storage Options, by model:
.-----+-----+-------------+----------------------------------. |9672/|# of | Memory | | | |Cards| | | | |Model| (1) | Min | Max | Increments | |-----+-----+------+------+----------------------------------| | RA6,|Config-same as RD6 | Increments- same for models RA6, | | R16,| 2 | " | " | R16, RB6, R26, RC6, RD6: 1GB, | | RB6,| 2 | " | " | 1.5 GB 2 GB, 2.5 GB, 3 GB, | | R26,| 2 | " | " | 3.5 GB, 4 GB, 5 GB, 6 GB, 7 GB, | | RC6,| 2 | " | " | 8 GB, 10 GB, 12 GB | | RD6 | 2 | 1 GB | 12 GB| | |-----+-----+------+------+----------------------------------| | R36 |Config- same as RX6| Increments- same for models R36, | | R46 | 4 | " | " | R46, R56, R66, R76, R86, R96, | | R56 | 4 | " | " | RX6, YX6: 2 GB, 2.5 GB 3 GB, | | R66 | 4 | " | " | 3.5 GB, 4 GB, 5 GB, 6 GB, 7 GB, | | R76 | 4 | " | " | 8 GB, 10 GB, 12 G B, 16 GB, | | R86 | 4 | " | " | 20 GB, 24 GB | | R96 | 4 | " | " | | | RX6 | 4 | 2 GB | 24 GB| | | YX6 | 4 | 8 GB | 24 GB| | |-----+-----+------+------+----------------------------------| | R06 | 2 or| 1 GB | 24 GB| Increments- 1 GB, 1.5 GB, 2 GB, | | | 4 | | | 2.5 GB, 3 GB, 3.5 GB, 4 GB, 5 GB,| | | | | | 6 GB, 7 GB,8 GB, 10 GB 12 GB, | | | | | | 16 GB, 20 GB, 24 GB | '-----+-----+------+------+----------------------------------'(1) Server storage, even though physically the same, can be configured as both Central and Expanded storage. All parts removed or replaced are the property of IBM and must be returned to IBM.
Parallel, ESCON, & FICON Channels (#2303, #2304, #2313, #2314)
Total Channels: Minimum -- 4; Maximum -- 256
Note: *Feature 2303 has increments of 3 parallel channels; feature 2304 has increments of 4 parallel channels. Feature 2303 will be As Available order status in 1999.
Coupling Links (#0007, #0008, #0216)
One HiPerLink adapter (#0216) is required for every two Coupling Links. The Intersystem Channel Adapter is not hot pluggable, but the Coupling Links are. Additional Intersystem Channel Adapters and Coupling Links can be ordered to reduce outages. Two types of Coupling Links are available on the G5 Models:
S/390 Open Systems Adapter 2
(#5201, #5202, #5206, #5207, #5208)
Cryptographic Coprocessor
Cryptographic Coprocessor is a data security standard feature on the S/390 G5 Parallel Enterprise Server servers. There are three levels of security which conform to export requirements. The Trusted Key Entry (TKE) feature is a workstation which provides secure Master Key access. This feature is not offered on the Coupling Facility R06 model; however, R06 coupling facility models upgraded to G5 server models are Cryptographic coprocessor-capable.
Feature
Description of Features Code
Cryptographic Coprocessor Hardware 0800
Feature
DES with PKA 0814
DES with PKA & TKE 0815
T-DES with PKA 0834
T-DES with PKA & TKE 0835
TKE Hardware (for Token Ring) 0806
Smart Card Reader 0807
TKE Hardware (for Ethernet) 0809
Hardware Management Console (HMC) with RSF (#0041)
The Hardware Management Console provides a single point of control and single system image for managing local or remote hardware elements such as 9672, 9674, 2003, and 3000 systems. The Hardware Management Console employs a state of the art direct-manipulation object-oriented graphical user interface supporting exception based real-time system status reporting via ICON's and colors, consolidated hardware messages, consolidated operating system messages, and consolidated service support (including modems and phone lines) and, of course, full operation of the managed systems. The Hardware Management Console(s) ordered with a S/390 G5 server are downward compatible with the Hardware Management Console(s) and SE(s) on the 9672 R1, R2, R3, G3, G4, G5 or 9674 C01, C02, C03, C04, C05, 2003, or 3000 systems. To allow the existing Hardware Management Console(s) on the 9672 R1, R2, R3, G3, G4, or 9674 C01, C02, C03, C04, C05, 2003, or 3000 systems to be upward compatible with the S/390 G5 Server an MES order will be required to convert features (#0009) or (#0021) to feature (#0031) or feature (#0041).
Optional System Assist Processor (SAP) (#0990)
S/390 G5 Parallel Enterprise Server I/O processing can be enhanced by taking advantage of the optional SAP features. These SAP features are available along with the default base SAP's that are assigned with each model. The base plus optional SAP's on a specific model must be equal to or less than the number of PU's. The configurator will only allow ordering of the maximum: models RA6, R16, RB6, R26, RC6, RD6, R36, R46, R56, R66, R76 and R06 all have (1) SAP standard; models R86, R96, RX6 and YX6 all have (2) SAPs standard. Use feature #0990 to order optional SAPs above the standard default base.
Power Sequence Control (PSC) (#6301 and #6302)
Provides power sequencing for attachment of up to 32 I/O controllers, if Internal Battery Feature (#2210) will be ordered. Feature #6301 provides a PSC box which can attach to 16 I/O controllers. Feature #6302 provides two PSC boxes which can attach to 32 I/O controllers. Feature #6301 and #6302 are mutually exclusive.
Expansion Cage (#2020, #2021-YX6 only)
This expansion cage provides space for additional channel cards in the G5 models for a maximum of 88 ESCON or 66 Parallel (feature #2303), or 88 Parallel channels (feature #2304) per cage.
Additional Frame (#3020)
The additional frame would be required when the second expansion cage is installed.
Internal Battery Feature (#2210)
The Internal Battery Feature (IBF) provides the function of a local Uninterruptible Power source. It has continuous self-testing capability for battery backup which has been fully integrated into the diagnostics, including Remote Service Facility (RSF) support. The IBF enables between 3.5 to a minimum of 20 minutes of full power hold-up for the 9672 Parallel Enterprise Server and up to one hour for the 9672 R06 Coupling Facility in power save mode. On upgraded single frame configurations, the IBF (#2210) and prior model power sequence controls, features (#6201 and #6202) are mutually exclusive. Refer to Power Sequence Control (#6301 and #6302).
Local Uninterruptible Power Supply (#9950)
To reduce system power outages, a local Uninterruptible Power Supply (UPS), Machine Type 9910, is available to be ordered with the new servers and coupling facilities announced today. This unit provides between 3.5 to 20 minutes of full power hold-up in the event of extended customer power line disturbances. Up to three additional battery modules may be ordered increasing the total holdup duration to between 10 minutes and 1 hour. The 9910 Local UPS is compatible with G5 models whose total Kilowatt load does not exceed 5 KW. refer to Hardware Announcement 196-068 dated April 2, 1996, for further details. The 9910 is fully compatible with the Internal Battery Feature (IBF). The 9910 Local UPS is compatible with G5 models whose total Kilowatt load does not exceed 5 KW.
External Time Reference (ETR) (#6150, #6152, #6153)
This feature provides the attachment of a Sysplex Timer (R) to set the Time-Of-Day (TOD) clocks in a multiprocessor environment and maintain synchronization. The ETR Master Card has one fiber optic port and one electrical port when used with the ETR External Cable (#6153). This feature is mutually exclusive with ETR Dual Port Card (#6152).
This feature provides the attachment of a Sysplex Timer to set the Time-Of-Day clocks in a multiprocessor environment and maintain synchronization. The Sysplex Timer attachment has two fiber optic ports allowing two sysplex timers to be attached, providing high availability. This provides redundant connectivity to the Sysplex Timer expanded availability configuration. This feature is mutually exclusive with ETR Master Card (#6150).
The ETR External Cable can be used to connect pairs of G5 models to enable the sharing of sysplex timer ports. ETR Master Card feature (#6150) must be ordered concurrently with this feature.
9032 Model 005 ESCON Director FICON Bridge card (#5260)
The 9032 Model 005 ESCON Director will be the first device to support FICON channels. Each 9032 Model 005 FICON Bridge card Feature (#5260) replaces one 8-port ESCON port card and is capable of performing the work of up to 8 ESCON channels. Up to 16 FICON Bridge cards will be supported in a single 9032 Model 005. This feature will provide customer investment protection by allowing currently installed ESCON control units to be shared between FICON and ESCON channels without modification.
9032 Model 005 ESCON Director FICON Enablement Feature (#5700)
The 9032 model 005 FICON Enablement Feature (#5700) provides power and logic upgrades necessary for installation of the FICON Bridge card feature (#5260). One feature (#5700) is required per 9032 model 005; installation is necessary with the first installation of a FICON Bridge card.
JCM OPS (#2029)
This feature is used for enabling the G5 to run the Japanese Operating System.
Booster Feature (#7998)
This feature is ordered to change the standard module to a 12 PU module and bring the total of memory cards to 4. It is orderable on model R16 only.
Booster Feature (#7999)
This feature is ordered to change the standard module to a 12 PU module and bring total of memory cards to 4. It is orderable on Model R26 only.
Maximum CP 12 Module (#7990)
This feature is ordered to increase the number of PU's in R36, R46, R56, and R66 models from 8 to 12. This may be desirable for future upgrades as well as situations where additional STIs are needed.
Software Requirements
The IBM 9672 G5 Parallel Enterprise Server models will be supported by the following operating systems in Basic mode and in LPAR mode:
Note: Use IBM Service Link or contact your IBM representative to obtain a current copy of the Preventive Service Planning (PSP) Bucket (UPGRADE: 9672DEVICE, SUBSET: 9672MVS/ESA). This contains specific MVS/ESA (TM) software planning information; for example, Authorized Program Analysis Reports (APARs) and Program Temporary Fixes (PTFs) required for each of the following levels of support.
Provides support for the Subspace Group Facility and growth to coupling-capable platforms. Refer to Software Announcement 294-152 dated April 6, 1994, for further details.
Provides support for ESCON channel-to-channel operated in Basic mode, the Asynchronous Pageout Facility, and Logical String Assist which were announced June 16, 1992. Refer to Hardware Announcement 192-146 dated June 6, 1992, for further details.
Also provides support for ESCON Multiple Image Facility (LPAR Mode), Hardware-Assisted Data Compression, and Asynchronous Data Mover Facility.
ICSF/MVS 2.1 is available at no charge to support the Cryptographic Coprocessor function on the new S/390 G5 Parallel Enterprise Server processors; it must be ordered as a standalone program product whenever OS/390 R1, R2, or R3 is the operating system ISCF/MVS 2.1 is included in the base for OS/390 R4 and R5.
The following VM releases are supported in Basic and LPAR mode:
There is a required maintenance level for VM/ESA to IPL on processors with the Cryptographic Coprocessor Feature. Existing VM/ESA systems should be upgraded to the required RSU level prior to the installation of a S/390 Server with the Cryptographic Coprocessor. The required RSU level which incorporates the PTFs mentioned above for each of the currently supported releases of VM/ESA are as follows:
VM/ESA Version 2 Release 3 does not require a specific service level.
The S/390 Generation 5 Enterprise Server models do not support S/370 (TM) mode operations. Therefor, the following VM systems or functions will not execute on a G5 server.
The following VSE versions are supported in Basic and LPAR mode:
Note: VSE/ESA Version 1 Release 4 continues to be available at this time and is best viewed as an accommodation which allows customers to begin the work needed to protect their extensive investments in VSE technology. Longer term, even VSE/ESA Version 1 Release 4 users should plan to move to an environment with VSE/ESA Version 2 Release 2 (or later) and S/390 servers.
The base product and the High Performance Option (HPO) Loosely Coupled facility supports the 9672 processors. When running in a loosely coupled environment, the Sysplex Timer attachment and TPF ESA mode is required. TPF ESA mode is standard on the 9672 processors. To achieve full performance in multi-engine environments, TPF ESA mode operation is required. TPF supports up to 8 PUs per TPF image; models R96, RX6 and YX6 are supported in I/O Intensive mode only.
Compatibility: No S/370 Mode Support
No S/370 mode guest support is provided on the S/390 G5 Enterprise Server processors. This applies to no S/370 mode LPAR support and no S/370 guest support under VM. VM customers should refer to the VM/ESA CP Programming Services guide regarding the 370 Accommodation Facility for programs that currently require running as a S/370 mode guest.
This product utilizes only existing attachment interfaces.
Any program written for ESA/370 (TM) architecture, or written to an optional ESA/390 (TM) architectural facility (ES Connection, Multisystems) when installed in ESA/370 mode, will run on a machine executing according to the ESA/390 architecture, provided that the program:
The CEC serial number (CPU identification number) is a portion of the information stored by the STORE CPU ID (STIDP) instruction. When an installed CMOS processor is upgraded to one of the processors announced today, the CEC serial number stored by STIDP on the upgraded processor will be the same as it was prior to the upgrade.
Planning Information
Customer Responsibilities: The customer is responsible for site preparation, scheduling of installation time, and ensuring that the prerequisite consoles and input/output devices are available. Physical planning assistance is available through your IBM representative.
Cable Orders: Line cords should be installed according to the national safety standards and will be provided based on the three digit country code. Refer to Installation Manual- Physical Planning, GC22-7102, for specifics.
Cables for ESCON channels are not provided with the channel features on any 9672 processors. Only nominal length 78D to OEMI parallel channel cables are provided with the parallel channel features on any 9672 processors. Planning the channel cable configuration, ordering, arranging delivery and installing prior to the 9672 installation is a customer responsibility. refer to I/O Equipment Installation Manual-Physical Planning, GC22-7064, for information on planning the channel cabling for specific control units.
Parallel channels attach via 78D cables alone or via 78D cables connected to OEMI bus and tag channel cables. Refer to the S/390 9672 Installation Manual-Physical planning, GC22-7102, for information.
Fiber optic jumper cables are required for connecting one Coupling Link feature to another Coupling Link feature. For initial equipment orders and feature MES orders, one standard-length IBM multi-mode or one standard-length single-mode fiber optic jumper cable will be provided at no additional charge for the connection between two Coupling Link features. The fiber optic jumper cables do not ship automatically with the Coupling Link features and must be ordered separately. Custom lengths will be provided at a charge. Refer to the Coupling Link Fiber Optic Cable section in the 9672/9674 Sales Manual pages.
The Hardware Management Console (HMC) is attached to the S/390 processor using a Token-Ring LAN attachment. A Token-Ring patch cable is shipped with the processor. Refer to the S/390 9672/9674 Installation Manual-Physical Planning, GC22-7102, for information on LAN connections required and addition LAN cable ordering information.
For S/390 Open Systems Adapter 2 cable planning information, refer to the 9672 Sales Manual pages.
For fiber optic components and connectivity solutions, contact your local IBM Marketing or Connectivity Service Representatives.
IBM's Fiber Transport Services (FTS)
IBM's Global Services FTS-III Direct Attach fiber optic trunk offering is now available for the 9672 G5 Models.
FTS-III Direct Attach fiber optic trunk cables support ESCON , Parallel Sysplex (R) and Open Systems Adapter-2 (OSA-2) fiber optic link connections. Direct Attach trunk cables allow trunk cable connections directly from a S/390 9672 processor, S/390 Coupling Facility 9674 and the 9032 ESCON Director Models 002, 003, and 005 into a main patch panel facility or directly to each other. Direct Attach trunk cables for the 9672 OSA fiber links allow trunk cable connections directly from the 9672 to a patch panel box located in the customer's wiring closet.
FTS-III Direct Attach fiber trunk cables and harnesses are the foundation for a highly organized, protected (eliminates the need for a fiber conveyance system), and easily movable fiber connectivity solution. By utilizing FTS-III structured cabling system, significant cost savings can be realized when relocating, reconfiguring, or adding additional hardware to the customer's ESCON, Parallel Sysplex and OSA installations.
The trunk cables are all plenum-rated OFNP (Optical Fiber Non-conductive Plenum) and are UL-910 approved. Charges are provided by Global Services on a per-contractual basis. Contact your Global Services representative for details. For more information, refer to Hardware Announcement 195-326 dated October 31, 1995, IBM Enhances Fiber Transport Services (FTS) and Services Announcement 696-025 dated September 10, 1996, Fiber Optic Trunk Cables for S/390 Processors and Coupling Facilities.
Information about the Direct Attach trunk cables can be found in the following publications which are available through IBM publications (PUBORDER):
Order
Title Number
Fiber Transport Services, G544-6253
Building a Foundation for
ESCON Flexibility
IBM Fiber Transport Services GA22-724
(FTS) Direct Attach Physical
and Configuration Planning
Installation Instructions for SA22-7229
Fiber Transport Services (FTS)
Direct Attach Trunking Systems
in 9672 Processors and 9674
Coupling Facilities
Fiber Quick Connect
IBM is now offering factory installation of direct-attach fiber optic trunking harnesses as an option with the G5 Servers. Direct-attach trunking eliminates the requirement for individual jumper cable connections to each ESCON or FICON channel, Coupling Link, HiPerlink, or Fast Ethernet, FDDI, or 155 ATM OSA-2 feature. Instead of up to 72 jumper cables exiting the server frame, one 72 channel trunk cable exits the G5 server. Fiber cable bulk is reduced both inside the server and under the raised floor.
With direct-attach trunking, individual channel ports can be quickly connected to a Main Distribution Facility (MDF) where the ports are arranged in patch panel boxes. Access to the I/O features in your S/390 G5 Server for configuration changes is no longer required. The server remains secure. Configuration changes are simplified and are made in one controlled area, the MDF, by authorized personnel moving short jumper cables between patch panels instead of moving long jumper cables under the raised floor. This reduces the scheduled outage times associated with moves, adds and changes.
IBM's structured fiber optic trunking system (Fiber Transport Services) utilizes the latest fiber optic technology to improve installation and configuration management. The Fiber Quick Connect feature enables the trunk harnesses to be installed at the factory so your server can be quickly connected to trunk cables pre-installed at your site. IBM's new MDF patch panels use the Small Form Factor (SFF) fiber optic connector called the DC (Dual Connection). This smaller sized duplex connector reduces the required MDF floor space by 50% when compared to current duplex connectors. The DC patch panels also enable your server CHPIDs to be arranged in sequential order with their CHPID addresses factory-labeled. Two DC patch panels support up to 256 CHPID addresses, so planning for future growth is made easy as well.
When the Fiber Quick Connect feature is selected for factory installation, the G5 server will arrive on site with the direct-attach trunk harnesses already installed. The trunks and patch panels can be installed prior to the arrival of the G5 Server and with the factory installation of the direct-attach harnesses, the on-site server install time can be reduced by up to 80%. As part of the factory installation package, documentation will be provided showing the CHPID layout and how the direct-attach harnesses are plugged.
IBM Global Services can provide the direct-attach trunk cables, patch panels, MDF hardware and planning and installation required to complete this total structured connectivity solution.
IBM Global Services Division -- Supplementary Services
IBM Global Services, with 110,000 professionals in 164 countries, is the world's largest and most versatile IT services provider. Its capabilities span a complete range, including product support and Global Network (TM) services, as well as consulting, outsourcing, systems integration, and education and training. IBM Global Services Product Support Services can assist in the installation, implementation, and/or integration of S/390 products, which can be incorporated into a total solution for customers with the following needs:
Operational Support Services for Parallel Sysplex Exploitation: Parallel Sysplex Exploitation Services enable the customer to exploit their established Parallel Sysplex environment. The services provide assistance in implementing new features of Parallel Sysplex or refining the customer's current installation. These enhancements will increase the customer's value and benefits derived from the environment. Each of the services provides the tasks and deliverables to assist the customer in implementing the specified component or enhancement to their Parallel Sysplex environment. A menu of potential services enables the customer to select those services that will maximize their Parallel Sysplex environment. This menu allows customers to customize the set of Exploitation Services to fit their environment and current implementation. The technical skills provided will guide and assist customers in implementing the tasks to achieve the objective of the service. Throughout the service, IBM experts provide skills transfer to the customer's technical professionals The menu of services include:
System Exploitation
Service Descriptions
This service configures the IBM OS/390 Firewall Technologies to allow internal, Web-enabled and File Transfer Protocol (FTP) -- enabled clients (already installed and located on the secure side of the network) to access Web servers and FTP servers located on the external network to internal resources.
Highlights:
Simplifying and enhancing migration
Now, demanding reconfigurations are made simple with the use of our quick-disconnect, high-density 12-fiber MTP terminated push-on connectors. MTP terminated trunks and harnesses provide fast and effective reconnection of processors and support devices without disturbing channel hardware configurations.
Helping you pull it all together
The field model upgrade time range, system downtime, is 2 to 8 hours, depending on the upgrade. These times assume that the prerequisite ECs and MESs have been installed prior to model upgrade.
Field upgrades require different lengths of time depending upon the customer configuration, the amount of work that can be accomplished in parallel at the customer location, and the method used to do the upgrade.
9672 Pre-Installation Configuration Service
Pre-Installation Configuration service is a no-charge optional manufacturing based service which reduces system installation time for factory ordered S/390 G5 Parallel Enterprise Server and Coupling Facilities. This service, available now, allows customers to receive their server with the Hardware Management Console (HMC) and Service Element (SE) customized to their specifications prior to shipment. Contact your IBM representative for further details.
IBM SecureWay (TM)
IBM SecureWay provides a common brand for IBM's broad portfolio of security offerings: hardware, software, consulting and services to help customers secure their information technology. Whether addressing an individual need or creating a total enterprise solution, IBM SecureWay offerings provide the expertise required to plan, design, implement and operate secure solutions for businesses. Additional information on these offerings can be found through the IBM I/T Security Home Page:
Security, Auditability, and Control
Reliability, Availability, and Serviceability
S/390 G5 Servers reduce downtime by using standard features that provide high levels of reliability, availability, and serviceability (RAS).
Reliability
The standard features that provide a high level of reliability include:
Availability
The standard features that provide a high level of availability include:
All S/390 G5 Servers are provided with an enhanced processor design. Each Central Processor contains dual Instruction/Execution Units, which operate simultaneously. Results are compared, and in the event of a miscompare, Instruction Retry is invoked. This design simplifies checking, and virtually eliminates CP failures due to soft errors.
Fault tolerant design allows hardware recovery to be performed, in most cases, totally transparent to customer operation and eliminates the need for a repair action, or defers a repair action to a convenient time scheduled by the customer.
Spare Processor Units (PUs) are provided, whenever the configuration allows, to maintain performance levels should an active central processor (CP), Integrated Coupling Facility (ICF) feature or System Assist Processor (SAP) fail on a G3, G4 or G5 Server models. The general philosophy is to provide a spare chip whenever possible; that is, when the number of CPs, ICFs and SAPs is less than the total number of processing units (PUs) available.
New on the S/390 G5 Servers is Transparent CP/ICF Sparing. When originally announced on the G3 models, CP sparing (ICF feature not available at G3 announce) required an IML of the machine to activate the spare CP. Later on the G3 models and on the G4 models, Concurrent CP/ICF Sparing was introduced. With Concurrent CP/ICF Sparing, an IML is not required to activate a spare PU. In shared LPAR mode, Concurrent CP/ICF Sparing is automatic and does not required operator intervention; however, in Basic Mode or with LPAR dedicated partitions, operator action is required to bring the new CP/ICF on-line (Operating System must support configuring the spare CP on-line). With the G5 Servers, CP/ICF Sparing is transparent in all modes of operation and requires no operator intervention to invoke a spare CP. Because there are no extra chips available on the RX6 and YX6, these models do not support CP Sparing. The R06 model supports sparing for all ICF features (including the model configured with 10 ICF features).
As a further enhancement, in most cases, the application that was running on the failed CP will be preserved and will continue processing on a new CP with no customer intervention required. See Enhanced Application Preservation below.
Dynamic recovery is provided for failure of the System Assist Processor (SAP). In the event of a SAP failure, if a spare Processor Unit (PU) is available, in most cases the spare PU will be dynamically activated as a new SAP. If there is no spare PU available, and the CPC has more than one Central Processor (CP), an active CP will be reassigned as a SAP. In either case, there is no customer intervention required. This capability eliminates an unplanned outage and permits a service action, if necessary, to be deferred to a more convenient time.
Application Preservation was introduced on the G4 Servers and is enhanced on the G5 models to provide more comprehensive application recovery should a CP fail. Application Preservation captures the machine state in the event of a CP failure and will, in most cases, switch processing to a spare PU or another active CP without customer intervention. On the G3 and G4 models, the uniprocessors were not covered by Application Preservation; G5 uniprocessor models will now recover work due to a CP failure, in most cases, on a spare PU using Application Preservation. The RX6 and YX6, although without CP Sparing, will attempt to recover an application that was running on a failed CP on another active CP in the configuration. This capability helps eliminate unplanned outages, eliminates customer intervention in the recovery process, and preserves the customer's application processing environment.
There are two Cryptographic Coprocesor Elements available in the G3, G4 and G5 Servers. On the G5 models, the availability has been enhanced by making both elements available on 2-way models and above. On G3 and G4 models, two cryptographic elements are available on the 5-way models and above with the 1-way through 4-way models having only a single cryptographic element. Recovery of a cryptographic coprocessor element is done by the operating system; i.e., the operating system reschedules and dispatches the failed instruction on the other cryptographic coprocessor element.
Another new availability feature on the G5 Servers is a second path from each cryptographic coprocessor element to a spare PU. Normally, each crypto element is configured to a primary CP. Should a primary CP fail, the alternate PU would transparently replace (spare) the failed primary CP, maintaining the crypto element's operation. However, an alternate PU is available only if that PU is not configured into the system as another CP, ICF or SAP. The two PUs associated with the alternate path from each crypto element are the last to be assigned as CPs, SAPS or ICFs. It should be noted that if a primary CP is not available at IML, the cryptographic element will be configured with its associated alternate PU.
Dynamic ICF Expansion is a function that allows an ICF logical partition to acquire additional processing power from the LPAR pool of shared general purpose central processors being used to execute production and/or test work on the system. This function is very useful when the ICF partition in some G5 general purpose models backs up another Coupling Facility. In this event, the ICF, using Dynamic ICF Expansion, can acquire additional processing capacity to handle the full Coupling Facility workload. Also, Dynamic ICF Expansion can be used to handle a peak workload situation when the ICF is being used as the regular coupling facility. This feature adds considerable flexibility to the configurability of the ICF in G5 general purpose models and optimizes the use of the processing power in the system.
The Dynamic ICF Expansion feature is available to every ICF Coupling Facility partition using coupling facility links. Each ICF coupling facility partition has its own ability to specify the number of ICF features that are dedicated to that partition and the amount of additional capability it can acquire. The tradeoff between using ICF features and the CPs in the LPAR shared pool is the exemption from software license fees.
Dynamic ICF expansion is not recommended for use in a logical coupling facility that uses ICMF.
Dynamic ICF Expansion is available on G5 general purpose models RA6, R16, RB6, R26, RC6, R36, R46, R56, R66, R76, and R86.
The Dynamic Coupling Facility (CF) Dispatching function helps enable continuous computing in the event of a coupling facility failure without requiring a standalone backup coupling facility. Enhanced dispatching algorithms enable you to define a backup coupling facility in a logical partition (LPAR) on your system. While this logical partition is in backup mode, although it is sharing resources with other LPARs running other active workload, it uses very little processor resource. When the backup CF becomes active, only the resource necessary to provide coupling is allocated.
Memory error checking and correction code detects and corrects single bit errors. Also, because of the memory structure design, errors due to a single memory chip failure are corrected.
Memory cards are equipped with spare memory chips. During normal operations, the system monitors and records accumulation of failing bits in memory chips that are corrected by Error Correction Code (ECC). Before a failure threshold is reached which could result in an uncorrectable error, the system invokes a spare memory chip in place of the one with the accumulated failing bits. This action may prevent an unscheduled outage for replacement of the memory card.
PR/SM LPAR storage reconfigurations can occur allowing nondisruptive add or removal to any partition with a cooperating guest.
This capability removes the restriction of storage reconfigurations only being possible from an adjacent and above logical partition.
Subsystem storage protection and subspace group facility support, for use with CICS/ESA (R), prevents application software from overwriting CICS system software, control blocks, and address spaces
Storage background scrubbing provides continuous monitoring of storage for the correction of detected faults before the storage is used.
In the event of a memory card failure, the system can be restarted with half of the original memory. Processing can be resumed until a replacement memory card is installed.
Dynamic I/O configuration enhances system availability by supporting the dynamic addition, removal, or modification of channel paths, control units, I/O devices, and I/O configuration definitions to both hardware and software without requiring a planned outage.
It is possible to concurrently add ESCON, Parallel and OSA channels and Coupling Facility (CF) Links provided an STI is installed and there are unused channel positions and a channel adapter card (for channels)/HiPerLink adapter card available. This capability may help eliminate an outage to upgrade the channel configuration.
In the event of a failure of a Memory Bus Adapter, in most cases, the system will continue to run with only the I/O connections associated with the failed MBA deconfigured. In a system configured for maximum availability, alternate paths will maintain access to critical I/O. This capability enables the system to run, partially degraded, until the part is replaced, restoring full capacity.
Concurrent channel maintenance allows replacement of a channel card without having to take the system or the channel group down. All CHPIDs on the affected card must be configured offline.
The power system offers dual primary (AC) power feeds. Each feed is electrically isolated and enables redundant power paths to each server. Customers may elect to provide a dual electrical service to the server, further minimizing any outage due to a single path power interruption.
The AC and DC power subsystems are designed with N+1 redundancy. Failure of a power thermal component does not cause a system outage.
S/390 G5 servers have error checking and correction of L2 and L3 storage. The Servers also have line delete for L1 and L2 Caches and Directories and line relocate for the Caches and L2 Directory. S/390 G5 servers perform ECC, background scrubbing and dynamic sparing of L3 & L4 memory, as well as ECC on the system memory buses.
ETR connection provided across clusters to enable the cluster degrade capability. The design includes both the full- and half- populated module cases.
Concurrent maintenance enables the replacement of failed units concurrently with system operation. This enhances the processor availability by eliminating the need for system outage to effect the repair. Concurrent maintenance capability exists for the following elements:
Concurrent Code Patch allows the activation of a patch concurrent with system operation thereby increasing the availability of the processor by reducing scheduled outage for LIC maintenance. This capability exists for code for the following elements:
The Integrated Battery Feature (IBF) provides backup input power. The feature is packaged internal to the machine. It can be used with a UPS to provide additional protection.
The Local Uninterruptible Power Supply, Machine Type 9910, can be installed as a supplemental or as an alternative to a central UPS to secure system availability.
Redundant Coupling Links and ICBs can be configured between a processor and the Coupling Facility. This potentially removes a single point of failure for the processor's data sharing capability in the Parallel Sysplex environment.
Serviceability
The standard features that provide a high level of serviceability include:
CHARGES
The charges provided in this announcement are suggested retail prices
for the U.S. only and are provided for your information only. Dealer
prices may vary, and prices may also vary by country. Prices are
subject to change without notice. For additional information and
current prices, contact your local IBM representative.
Product Charges: Product structure will be available when the S/390 Parallel Enterprise Server -- Generation 5 models are announced at the end of second quarter 1998.
Mandatory Feature Reference Guide -- 9672 G5 Models
Feature Model
Description Number Note RA6 R16 RB6 R26 RC6 RD6 R36 R46
RA6 1-Way Processor 0901 (1) X
R16 1-Way Processor 0902 (1) X
RB6 2-Way Processor 0903 (1) X
R26 2-Way Processor 0908 (1) X
RC6 3-Way Processor 0904 (1) X
R36 3-Way Processor 0905 (1) X
RD6 4-Way Processor 0906 (1) X
R46 4-Way Processor 0907 (1) X
Storage
1.0 GB Memory 7010 (1) X X X X X X NA NA
1.5 GB Memory 7015 (1) X X X X X X NA NA
2.0 GB Memory 7020 (1) X X X X X X X X
2.5 GB Memory 7025 (1) X X X X X X X X
3.0 GB Memory 7030 (1) X X X X X X X X
3.5 GB Memory 7035 (1) X X X X X X X X
4.0 GB Memory 7040 (1) X X X X X X X X
5.0 GB Memory 7050 (1) X X X X X X X X
6.0 GB Memory 7060 (1) X X X X X X X X
7.0 GB Memory 7070 (1) X X X X X X X X
8.0 GB Memory 7080 (1) X X X X X X X X
10 GB Memory 7100 (1) X X X X X X X X
12 GB Memory 7120 (1) X X X X X X X X
16 GB Memory 7160 (1) NA X(3)NA X(3) NA NA X X
20 GB Memory 7200 (1) NA X(3)NA X(3) NA NA X X
24 GB Memory 7240 (1) NA X(3)NA X(3) NA NA X X
Channels
Parallel Channel Cd 2303 (1,2) X X X X X X X
Parallel Channel Cd 2304 (1,2) X X X X X X X
ESCON Channel Cd 2313 X X X X X X X
FICON Channel Cd 2314 X X X X X X X
OSA2 ETHRNT/TKEN Rin5201 X X X X X X X
OSA2 FDDI 5202 X X X X X X X
OSA2 ATM 155 Mb MM 5206 X X X X X X X
OSA2 ATM 155 Mb SM 5207 X X X X X X X
OSA2 FAST ETHERNET 5208 X X X X X X X
HiPerlinks 0216 X X X X X X X
Coupling Link 10 km 0008 X X X X X X X
Note: Model configurations require an MCM and a mandatory minimum of memory, channels and OSA-2.
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
(1) Parts removed or replaced become the property of IBM and must be returned.
(2) Feature #2303 (3 port Parallel Cards) is compatible with Feature #2304, and may be carried to G5 with an upgrade MES or installed with new builds in 1998. Feature #2303 will be "As Available" in 1999.
(3) ONE of the following three options must be satisfied:
Mandatory Feature Reference Guide -- 9672 G5 Models (cont'd)
Feature Model
Description Number Note R56 R66 R76 R86 R96 RX6 YX6
R56 5-Way Processor 0909 (1) X
R66 6-Way Processor 0910 (1) X
R76 7-Way Processor 0911 (1) X
R86 8-Way Processor 0912 (1) X
R96 9-Way Processor 0913 (1) X
RX6 10-Way Processor0914 (1) X
YX6 10-Way Processor0915 (1) X
Storage
1.0 GB Memory 7010 (1) NA NA NA NA NA NA NA
1.5 GB Memory 7015 (1) NA NA NA NA NA NA NA
2.0 GB Memory 7020 (1) X X X X X X NA
2.5 GB Memory 7025 (1) X X X X X X NA
3.0 GB Memory 7030 (1) X X X X X X NA
3.5 GB Memory 7035 (1) X X X X X X NA
4.0 GB Memory 7040 (1) X X X X X X NA
5.0 GB Memory 7050 (1) X X X X X X NA
6.0 GB Memory 7060 (1) X X X X X X NA
7.0 GB Memory 7070 (1) X X X X X X NA
8.0 GB Memory 7080 (1) X X X X X X X
10 GB Memory 7100 (1) X X X X X X X
12 GB Memory 7120 (1) X X X X X X X
16 GB Memory 7160 (1) X X X X X X X
20 GB Memory 7200 (1) X X X X X X X
24 GB Memory 7240 (1) X X X X X X X
Channels
Parallel Channel Cd 2303 (1,2) X X X X X X X
Parallel Channel Cd 2304 (1,2) X X X X X X X
ESCON Channel Cd 2313 X X X X X X X
FICON Channel Cd 2314 X X X X X X X
OSA2 EN/TR 5201 X X X X X X X
OSA2 FDDI 5202 X X X X X X X
OSA2 ATM 155 Mb MM 5206 X X X X X X X
OSA2 ATM 155 Mb SM 5207 X X X X X X X
OSA2 FAST ETHERNET 5208 X X X X X X X
HiPerlinks 0216 X X X X X X X
Coupling Link 10 km 0008 X X X X X X X
Note: Model configurations require an MCM and a mandatory minimum of memory, channels and OSA.
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
(1) Parts removed or replaced become the property of IBM and must be returned.
(2) Feature #2303 (3 port Parallel Cards) is compatible with Feature #2304, and may be carried to G5 with an upgrade MES or installed with new builds in 1998. Feature #2303 will be "As Available" in 1999.
Mandatory Feature Reference Guide -- 9672 R06 Model
Feature R06: # of ICFS
Description Number Note 1 2 3 4 5 6 7 8 9 10
1-ICFs 7971 (1) X
2-ICFs 7972 (1) X
3-ICFs 7973 (1) X
4-ICFs 7974 (1) X
5-ICFs 7975 (1) X
6-ICFs 7976 (1) X
7-ICFs 7977 (1) X
8-ICFs 7978 (1) X
9-ICFs 7979 (1) X
10-ICFs 7980 (1) X
Storage
1.0 GB Memory 7010 (1) X X X X NA NA NA NA NA NA
1.5 GB Memory 7015 (1) X X X X NA NA NA NA NA NA
2.0 GB Memory 7020 (1) X X X X X X X X X X
2.5 GB Memory 7025 (1) X X X X X X X X X X
3.0 GB Memory 7030 (1) X X X X X X X X X X
3.5 GB Memory 7035 (1) X X X X X X X X X X
4.0 GB Memory 7040 (1) X X X X X X X X X X
5.0 GB Memory 7050 (1) X X X X X X X X X X
6.0 GB Memory 7060 (1) X X X X X X X X X X
7.0 GB Memory 7070 (1) X X X X X X X X X X
8.0 GB Memory 7080 (1) X X X X X X X X X X
10 GB Memory 7100 (1) X X X X X X X X X X
12 GB Memory 7120 (1) X X X X X X X X X X
16 GB Memory 7160 (1) NA NA NA NA X X X X X X
20 GB Memory 7200 (1) NA NA NA NA X X X X X X
24 GB Memory 7240 (1) NA NA NA NA X X X X X X
HiPerlinks 0216 X X X X X X X X X X
Coupling Link 10008 X X X X X X X X X X
Integrated Clus0992Bus X X X X X X X X X X
Note: Model configurations require an MCM, and a minimum of memory, and a coupling connection (HiPerlinks or ICBs).
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
(1) Parts removed or replaced become the property of IBM and must be returned.
Optional Feature Reference Guide -- 9672 Models (NEW BUILD)
Feature Models
Description Number Note 9672 G5* 9672 R06* *
Coupling Link 10 km 0008 All All
HiPerlinks (Coupling Links) 0216 All All
Token-ring/HMC 0023 All All
3270/HMC 0026 All All
Ethernet/SE 0034 All All
HMC Console with RSF 0041 All All
Sml Console Display 6090 All All
Lrg Console Display 6091 All All
ICF (Internal Coupling Fac.) 0991 (7) See Note All
ICB (Integrated Cluster Bus) 0992 All All
TKE Hardware (for Token Ring) 0806 All NA
Smart Card 0807 All NA
TKE Hardware (for Ethernet) 0809 All NA
I/O Expansion Cage 2020 All All
I/O Expansion Cage-YX6 2021 (6) See Note NA
JCM OPS 2029 All NA
Internal Battery Feature 2210 All All
Fiber Quick Connect 2300 All All
FICON Channel Cd 2314 All NA
Additional Frame 3020 All NA
ETR Master Cd 6150 All NA
ETR Dual Port 6152 All NA
ETR Ext Cable 6153 All NA
1 Pwr Ctlr 6301 All NA
2 Pwr Ctlrs 6302 All NA
Maximum CP (12 PU) Module 7990 (8) See Note See Note
Booster Feature (Requires RPQ)7998 (9) See Note NA
Booster Feature (Requires RPQ)7999 (10) See Note NA
Note: Hardware prices for the features of the 9672 models are available from your IBM representative.
Note: Use this notice as the source for previously announced features.
(*) 9672 Models RA6, R16, RB6, R26, RC6, R36, RD6, R46, R56, R66, R76, R86, R96, RX6 and YX6.
(**) 9672 Model R06 only.
(3) Export License Mandatory
(4) Reserved for announcement second quarter 1998
(5) Reserved for announcement second quarter 1998
(6) Model YX6 only
(7) Models RA6, R16, RB6, R26, RC6, R36, R46, R56, R66, R76, R86 only; Models RD6 and R96 will also be able to have an ICF effective March 31, 1999.
(8) Models R16, R26, R36, R46, R56, R66 only; model R06 1, 2, 3, 4 ICFs only.
(9) Model R16 only with RPQ # 8P1959 only
(10) Model R26 only with RPQ # 8P1960 only
Optional Feature Reference Guide -- 9032 -- 005 ESCON Director
Feature
Description Number Note
FICON Bridge Card 5260 (1)
FICON Enablement feature 5700 (1)
(1) This feature is orderable beginning 4Q 1998
Coupling Link Fiber Optic Cable Groups -- 9672 Models
Cable
Assembly Models
Description Number 9672 G5* 9672 R06
Singlemode Cable #8311 08H2774 (See note All
(*) 9672 Models RA6, R16, RB6, R26, RC6, R36, RD6, R46, R56, R66 R76, R86, R96, RX6, YX6. 9672 Model R06.
FICON Link Fiber Optic Cable Groups -- 9672 Models
Cable
Assembly Models
Description Number 9672 G5
Singlemode Cable #8311 08H2774 All
Optional SAP Feature Reference Guide -- 9672 G5 Models
OrderFeature Model
Description QuantNumber RA6 R16 RB6 R26 RC6 R36 RD6 R46
1-SAP (1) 0990 NA NA X X X X NA X
2-SAP (2) 0990 NA NA NA NA NA X NA X
3-SAP (3) 0990 NA NA NA NA NA NA NA X
4-SAP (4) 0990 NA NA NA NA NA NA NA NA
5-SAP (5) 0990 NA NA NA NA NA NA NA NA
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
Note: SAPs are ordered by FC 0990 and quantity (N). Models RD6 and R96 will be able to be ordered with an additional SAP March 31, 1999.
Optional SAP Feature Reference Guide -- 9672 G5 Models (cont'd)
OrderFeature Model
Description QuantNumber Note R56 R66 R76 R86 R96 RX6 YX6
1-SAP (1) 0990 X X X X NA NA NA
2-SAP (2) 0990 X X X NA NA NA NA
3-SAP (3) 0990 X X X NA NA NA NA
4-SAP (4) 0990 X X NA NA NA NA NA
5-SAP (5) 0990 NA NA NA NA NA NA NA
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
Note: SAPs are ordered by FC 0990 and quantity (N). Models RD6 and R96 will be able to be ordered with an additional SAP March 31, 1999.
Maximum Optional ICF Feature Reference Guide -- 9672 G5 Models
Order Feature Model
Description Quantity NumberR06 RA6 R16 RB6 R26 RC6 R36 RD6 R46
1-ICF (1) 0991 X X X X X X X NA X
2-ICF (2) 0991 X X X X X NA X NA X
3-ICF (3) 0991 X X X NA NA NA X NA X
4-ICF (4) 0991 X NA NA NA NA NA X NA X
5-ICF (5) 0991 X NA NA NA NA NA X NA X
6-ICF (6) 0991 X NA NA NA NA NA X NA X
7-ICF (7) 0991 X NA NA NA NA NA X NA NA
8-ICF (8) 0991 X NA NA NA NA NA NA NA NA
9-ICF (9) 0991 X NA NA NA NA NA NA NA NA
10-ICF (10) 0991 X NA NA NA NA NA NA NA NA
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
Note: ICFs are ordered by FC 0991 and a specified quantity. Feature #7971 through #7980 are identified with R06 New Builds and Upgrades. In the event that a Model R06 is converted to any Model RA6 through R96, features #7971 through #7980 will be converted to feature #0991. Effective March 31, 1999, the maximum ICF features will be increased by 1, with the use of the last spare; Models RD6 and R96 will then be able to have one ICF.
Max. Optional ICF Feature Reference Guide -- 9672 G5 Models (cont'd)
Order Feature Model
Description QuantityNumber R56 R66 R76 R86 R96 RX6 YX6
1-ICF (1) 0991 X X X X NA NA NA
2-ICF (2) 0991 X X X NA NA NA NA
3-ICF (3) 0991 X X X NA NA NA NA
4-ICF (4) 0991 X X NA NA NA NA NA
5-ICF (5) 0991 X NA NA NA NA NA NA
Note: Use this notice as the source for previously announced features. All features are initial and MES orderable.
Note: ICFs are ordered by FC 0991 and a specified quantity. Feature #7971 through #7980 are identified with R06 New Builds and Upgrades. In the event that a Model R06 is converted to any Model RA6 through R96, features #7971 through #7980 will be converted to feature #0991. Effective March 31, 1999, the maximum ICF features will be increased by 1, with the use of the last spare; Models RD6 and RX6 will then be able to have one ICF.
Specify Codes 9672 G5 Models
Feature Models
Description Number Code Note 9672 G5* 9672 R06**
CEC/Cage Airflow Cd 0012 All All
ISC Airflow Cd 0013 All All
Channel Driver Cd (5L CHA) 0018 All NA
CEC Cage 0057 (1) See Note All
CEC Cage 0058 (2) See Note NA
CEC Cage 0059 (5) See Note NA
STI Extender Card 0994 All All
CRYPTO 0800 All NA
No Crypto 0808 All NA
CDMF w/Exportable PKA 0811 All NA
DES w/Exportable PKA 0812 (3) All NA
DES w/Exportable PKA & TKE 0813 (3) All NA
T-DES w/Exportable PKA 0832 (3)(6) All NA
T-DES w/Exportable PKA & TKE 0833 (3)(6) All NA
DES w/PKA 0814 All NA
DES w/PKA & TKE 0815 All NA
Feature Models
Description Number Code Note 9672 G5* 9672 R06**
T-DES w/PKA 0834 (6) All NA
T-DES w/PKA & TKE 0835 (6) All NA
Custom Quick Shipment 1748 All All
FIBB Card-Single Wide 2339 All All
Language Grp-Portugese 2978 (3) All All
Language Grp-China 5562 All All
4.8 Non-US 8888 All All
Northern Hemisphere 9930 All All
Southern Hemisphere 9931 All All
UPS 9910 9950 All All
MCM Service Tool Kit 9962 All NA
Parallel Sysplex Offering 9970 All All
Frame Reduction for shipping 9978 (5) See Note NA
Note: Use this notice as the source for previously announced features.
(*) 9672 Models RA6, R16, RB6, R26, RC6, RD6, R36, R46, R56, R66, R76 R86, R96, RX6 and YX6.
(**) 9672 Model R06 only.
(1) 9672 Models RA6, R16, RB6, R26, RC6, RD6; Model R06 with 1-4 ICFs.
(2) 9672 Models R16, R26, R36, R46, R56, R66, R76, R86, R96, RX6: Model R06 with 5-10 ICFs.
(3) Canada, AP, LA and EMEA only.
(4) Saudia Arabia only.
(5) Model YX6 only.
(6) T-DES (Triple DES Encryption exports to World Trade countries (except Canada)) require special authorization. Contact your IBM Representative.
Model Conversions
Model conversions will be announced at Product Announcement second quarter 1998.
For all local charges, contact your IBM Representative.
Trademarks
PR/SM, MVS, OS/390, MVS/ESA, S/370, VSE/ESA, ESA/370, ESA/390,
System/370, System/360, Processor Resource/Systems Manager,
Global Network, SmoothStart, IMS, and SecureWay are trademarks
of International Business Machines Corporation in the United
States or other countries or both.
S/390, CICS, ESCON, DB2, PS/2, System/390, OS/2, NetView,
RETAIN, Sysplex Timer, VM/ESA, Parallel Sysplex, VTAM,
OfficeVision, and CICS/ESA are registered trademarks of
International Business Machines Corporation in the United
States or other countries or both.
UNIX is a registered trademark in the United States and other
countries exclusively through X/Open Company Limited.
Domino is a trademark of Lotus Development Corporation.
Lotus is a registered trademark of Lotus Development
Corporation.
Other company, product, and service names may be trademarks or
service marks of others.