IBM DB2 Analytics Accelerator for z/OS V4.1.0: Release Notes
The release notes contain the latest information about IBM DB2 Analytics Accelerator for z/OS V4.1.0.
(1) What's new
This version of IBM DB2 Analytics Accelerator for z/OS, when used with the latest
DB2 for z/OS support, includes the following new features:
- Support for acceleration of static SQL queries in DB2 for z/OS applications.
- DB2 for z/OS multi-row fetching of results from accelerated queries (a
configurable number of rows can be retrieved in a single time unit)
- Workload balancing of accelerated queries. If a query can be executed on more
than one accelerator, DB2 for z/OS automatically distributes the workload
among qualifying accelerators. Accelerators with a low utilization are favored. This feature includes fail-over support: If one accelerator becomes unavailable, the workload is rerouted to available accelerators.
- Support for multiple encoding. If tables in the same DB2 subsystem are encoding in EBCDIC and UNICODE, they can now be defined on the same accelerator for query acceleration.
- High-performance storage saver (HPSS) improvements. A full-scale restore
function has been added to the HPSS scope. This function takes advantage of the
newest DB2 for z/OS Utilities support. In addition, an enhanced DB2 for z/OS
Utility function now sets the affected table space to a persistent read-only state
when the HPSS is invoked to move partition data. Furthermore, the same
partition data can now be moved to multiple accelerators.
- Table-load performance improvements, resulting in a two-digit percentage of
- Performance improvement for incremental updates.
- Incremental updates configurable for up to ten DB2 subsystems.
- Automated transfer and installation (apply action) of updates for the Netezza
Performance Server® (NPS®)
- Incremental updates continue while tables are being reloaded.
- Workload Manager (WLM) support for local DB2 for z/OS applications. That is,
WLM priorities can be assigned to local applications to support your processing
- Support for new features and functions DB2 11 for z/OS
- Expanded information for monitoring applications
(2) Multiple Encoding considerations
The new AQT_ENABLE_MULTIPLE_ENCODING environment variable allows you to define EBCDIC and UNICODE tables that are located in the same DB2 subsystem on the same accelerator. However, using this configuration might result in sorting differences between DB2 for z/OS and the accelerator for certain queries on UNICODE tables. See the documentation for AQT_ENABLE_MULTIPLE_ENCODING in the IBM DB2 Analytics Accelerator for z/OS: Stored Procedures Reference.
If you want to define both, UNICODE and EBCDIC tables that are located in the same DB2 subsystem on the same accelerator, specify the AQT_ENABLE_MULTIPLE_ENCODING environment variable before adding the first table (from the subsytem) to the accelerator.
(3) Avoid locking problems between DB2 11 automatic index cleanup processing and loading replication-enabled table to IBM DB2 Analytics Accelerator for z/OS
Applies to: all DB2 11 users who
- Use the incremental update feature of IBM DB2 Analytics Accelerator for z/OS
- Work with a non-zero value of the INDEX_CLEANUP_THREADS subsystem parameter.
When loading a table that is enabled for replication in DB2 11, index cleanup processing may take place at that time.
There are situations, where following error might be returned by the ACCEL_LOAD_TABLES stored procedure in the MESSAGE output parameter:
<?xml version="1.0" encoding="UTF-8" ?>
<dwa:messageOutput xmlns:dwa="http://www.ibm.com/xmlns/prod/dwa/2011" version="1.0">
<message severity="error" reason-code="AQT20021I">
<text>A table could be loaded successfully, but a subsequent incremental update operation on this table failed. Message: "tion "eReplication" failed: "CHC0102E SQL error:
DSNT408I SQLCODE = -911, ERROR: THE CURRENT UNIT OF WORK HAS BEEN ROLLED BACK DUE TO
DEADLOCK OR TIMEOUT. REASON 00C9008E, TYPE OF RESOURCE 00000100, AND RESOURCE
DSNT418I SQLSTATE = 40001 SQLSTATE RETURN CODE
DSNT415I SQLERRP = DSNXIATB SQL PROCEDURE DETECTING ERROR
DSNT416I SQLERRD = -190 0 0 -1 0 0 SQL DIAGNOSTIC INFORMATION
DSNT416I SQLERRD = X'FFFFFF42' X'00000000' X'00000000' X'FFFFFFFF' X'00000000'
X'00000000' SQL DIAGNOSTIC INFORMATION, Diagnostic information (paaAddTableToCatalog;14,532)""
<description>An incremental update operation failed after a successful load operation, probably because incremental updates were not enabled for the table. However, the table remains in LOADED state, which means that it can be used for accelerated queries.
<action>Try to enable incremental updates for the table.
To ensure that no DB2 index cleanup processes are running at a time when tables are loaded, follow the recommendations in the chapter "Controlling index cleanup processing" of the the DB2 11 for z/OS documentation:
(4) Availability of "Workload Balancing" function and "expanded information for monitoring applications"
The new V4.1 "Workload Balancing" function and the "expanded information for monitoring applications" will be delivered in a separate DB2 11 PTF.
See Prerequisites and Maintenance for IBM DB2 Analytics Accelerator for z/OS Version 4.1.: http://www.ibm.com/support/docview.wss?uid=swg27039487
(5) SPU failover and disk replacement
You can now initiate the failover of a Netezza snippet processing unit (SPU) and a disk replacement
from the IBM DB2 Analytics Accelerator Console. See the following technotes:
Initiating a SPU failover:
Replacing a disk:
(6) FDT and HPF prerequisite checks for a migration from version 3 to version 4.
Migrating from version 3 to version 4, you are required to upgrade your current Netezza Performance Server (NPS) to NPS 7.0.2 P9.
However, before a new NPS is activated (apply step in IBM DB2 Analytics Accelerator Studio), it is checked whether the version of the NPS is compatible with the installed versions of the Firmware Diagnostics and Tools (FDT) and the Netezza Host Platform (HPF) on the accelerator. The required minimum HPF level is 5.3.1.
If the FDT or the HPS is incompatible, you must update these components to a level that is supported by the certified NPS driver. For these updates, you need help from IBM support. An update of the FDT or the HPF by IBM support takes between 8 and 12 hours. Consider this as you work on your migration schedule. A service window of 8 hours is required in most cases. However, an update of a Netezza Twin Fin 48 or an even more powerful machine requires at least 12 hours.
Therefore, do not start the update right away. Instead, proceed as follows:
- Create a trace file that contains the relevant system information. A trace level of DEFAULT is sufficient for this.
- Open a service request (PMR), preferably two weeks before the planned migration date. This gives IBM support enough time to analyze your system, coordinate maintenance actions, and update components as required.
- State your intent to migrate to version 4 in the body of the PMR
- Attach the trace file to the request.
IBM support will analyze the contents of the trace file, compare the current FDT and HPF levels of the machine with the required minimum level for NPS 7.0.2.P9 and, if necessary, coordinate a system upgrade of the components that do not fulfill the requirements.
After installing the PTF that contains the certified driver and transferring the corresponding software package, you can upgrade the NPS yourself by using the Transfer Software and Apply Software functions in IBM DB2 Analytics Accelerator Studio. This update takes about 1 hour.
Here is a summary of the estimated service times, broken down by component:
NPS: 1 hour
HPF: 1 hour
FDT: 6 -10 hours
See how much time the individual FDT components claim for an update:
AMM - 1-3 hours
Blades - 2-6 hours
Storage Media - 30 min. - 1 hour.
Host Bios Firmware - 2 hours (1 hour for each host) - sometimes 3 hours (1.5 hours per host)
HBA - 15 min.
Note that the required upgrade time varies from system to system. It is influenced by the following factors:
- The current and the target levels of the NPS, the FDT, and the HPF
- Whether the system was recently upgraded
- Size of the system - sometimes blade updates take longer on big systems
- Hardware or platform issues during the upgrade. The system is scanned for issues before the upgrade, but this does not guarantee that no issues will be found during the upgrade.
The time estimates are based on a typical upgrade. There is no guarantee that the upgrade will not take longer than anticipated.
Translate this page: