IBM Solution Tracks and Enhances Network Performance

Solving problems through superior insight

Tivoli Beat - A weekly IBM service management perspective. Good network management requires good information. Establishing and quantifying just what is happening on a network is an essential part of supporting the complete range of services that fulfill user and customer needs.

Complex networks, however, are comprised of, or make use of, a vast array of related systems, services, applications, domains, network protocols and other elements. Obtaining quantified information pertaining to the performance of any logical subset of such a network is therefore no simple feat—and without that information, it is very difficult to achieve holistic network performance management.

The consequences can be both significant and far-reaching. When network problems occur, isolating their root cause on the network will be slower. The business impact those problems create will, in turn, be larger. The customer or end user experience will diminish, leading to a loss of productivity, revenue or even brand strength. None of these outcomes are desirable, to say the least.

Challenges of this type are usually associated with the communications service market, and for good reason. Most communications service providers (CSPs) have extraordinarily complex infrastructures—both wireless and wireline—and extraordinarily high expectations on the part of their users. However, the same challenges increasingly apply to the enterprise market as well. As enterprise networks become more solution-rich and complex, and the enterprise becomes more dependent on IT services for even the most basic productivity, these networks, too, will require a new, more sophisticated, more centralized approach to performance management. This is particularly true for enterprises in the financial sector, such as banks, for whom the highest levels of service availability and network performance are more than targets—they're de facto requirements.

IBM Tivoli Netcool Performance Analyzer delivers next-generation network profiling—today

“As enterprise networks become more solution-rich and complex, and the enterprise becomes more dependent on IT services for even the most basic productivity, these networks, too, will require a new, more sophisticated, more centralized approach to performance management.” IBM's response? IBM Tivoli Netcool Performance Flow Analyzer (TNPFA)—a new tool to obtain end-to-end resource usage information relating to every critical element of even the most complex network infrastructure, including applications, servers, services, protocols, domains, interfaces and various combinations of those elements.

Proactive network availability management
TNPFA gives network managers a detailed understanding of application traffic patterns, prevents/minimizes adverse effects and enables fast resolutions should any occur. The information it generates thus facilitates the planning, deployment and validation of new or upgraded network based service as well.

Reduced costs and higher efficiency

TNPFA yields detailed visibility of network traffic behavior. This, in turn, enables network operators to quickly identify and resolve the root causes of network performance problems – diminishing their business impact and minimizing their costs.

Rapid time to value

Thanks to its small footprint and straightforward installation, TNPFA starts generating business value almost immediately, yet requires relatively few resources for itself. And when used in combination with complementary solutions such as IBM Tivoli Network Performance Manager, IBM Tivoli Monitoring and IBM Tivoli Composite Application Manager, TNPFA can play a key role in achieving optimized network management.

A powerful solution with many strengths

How does TNPFA work? Deployment is simple; it is a lightweight solution with a small operational footprint, available for several Linux flavors, as well as AIX in both 32-bit and 64-bit versions. This design essentially guarantees almost any customer will already have suitable infrastructure to support it.

Following deployment, TNPFA monitors information streams on the network in real time. Thanks to its exceptional performance levels, it can support flow rates as high or higher than 10,000 flows per second—better performance than any competitive offering.

Part of this performance level comes as a consequence of fact that TNPFA can take advantage of parallel processing on multi-core and multi-processor systems; distributed installations can yield exceptional performance rates of 50,000 flows per second or more. Also relevant is the fact that TNPFA leverages a next-generation aggregation database capable of storing very large amounts of network data in very short amounts of time and on relatively small amounts of storage. The database's exceptionally efficient memory/disk synchronization means that data import and export times are minimized as well.

Such performance levels translate naturally into best-in-class scalability and flexibility. Organizations can leverage TNPFA to profile network performance both today and tomorrow, knowing that as their network traffic grows, TNPFA can quickly, easily, and cost-effectively be configured to scale up in proportion. And because TNPFA efficiency gives administrators exceptionally tight control over resource usage, generating the real-time information they need to optimize and protect the operation of business-critical networked infrastructures.

Furthermore, the range of network flow types the solution supports is unusually broad, complying with most of the major flow formats used in complex networks today. In addition to NetFlow (the network protocol used by Cisco Systems for its IOS devices) versions 5, 7 and 9, TFPFA also supports flows based on jFlow (Juniper Networks), cflowd (Alcatel/Lucent) and Huawei NetStream. While some network profilers only support IP version four, TNPFA supports IP version 6 at the data and control plane.

TNPFA even supports IPFIX, an export format designed for universal use and based on NetFlow v9; TNPFA has successfully passed IPFIX interoperability tests ensuring that IBM Tivoli customers can deploy it with confidence it will work on their networks with all IPFIX-capable devices. This is true of no other flow analysis and profiling tool available today.

Powerful analysis and reporting simplifies troubleshooting and guides network planning

Once TNPFA has obtained profiling information, it can then analyze the data, generating reports that fulfill any given business requirement. To access reports, administrators can create role-based users with different privileges or rights via the Web-based interface, which is accessible from any standard browser on the network. And once logged in, these users can then (depending on privileges) examine the results in a configurable Web environment that corresponds to their job duties.

The analysis TNPFA generates is similarly straightforward and convenient. Results can be depicted in tables, pie charts, graphs and other formats, and these are in turn "drillable"—if more detail is required within a given element, such as sorting criteria, time duration or packet or flow type, the user can easily obtain it.

TNPFA can also be configured to generate reports automatically, on any given time interval and in various formats (PDF, XML, JSON and text). They will show, as required, information about any subcategory or logical group of network flow data, including applications, hosts and servers, domains/subnets, individual sessions or end-to-end flows, traffic types, service types, protocols, TCP/UDP ports—or any combination of them.

For instance, it would be possible to create a report reflecting quality of service for a particular application as experienced by users on a given subnet. Such detail, available in whatever depth is dictated by business requirements, will translate into a clear picture of what is taking place on the network, as experienced by any given group of people using any given service for any given business purpose.

In cases of network congestion, resulting in diminished service levels, the causes of that congestion can thus be established clearly and easily. Accurate network forecasting, too, is enabled by TNPFA. Consider a situation in which two or more networks or servers will soon be consolidating; administrators will want the most accurate possible prediction on how well the consolidated infrastructure will support necessary workloads.

Thanks to TNPFA's comprehensive reporting, they can obtain this information in advance. TNPFA can serve as an oracle of expected future performance, leading to improved network planning in general by helping administrators anticipate unseen problems and work around them proactively.