Yankee lays out performance-optimization options

Last time, I discussed a Yankee Group report recently released that detailed why bandwidth doesn’t always solve network and application performance problems. This time, we’ll look at possible fixes.

Part of the challenge is that most network managers have a limited view into the applications running on the network, how those apps use bandwidth, and how that use affects overall performance.

To address the challenge, Yankee says there are two potential performance-optimization approaches: infrastructure-focused and management-focused. According to the research firm, most enterprises require a combination of the two to ensure that network devices and applications perform up to par.

The old standby, scaling up bandwidth, processing power or both, falls under the infrastructure-focused umbrella. The goal of this approach is to eliminate congestion by provisioning enough capacity or adding enough bandwidth to handle several application performance scenarios. A bonus to this approach is that it provides redundancy for companies seeking disaster recovery.

Yet there are some drawbacks. Yankee says the major drawbacks to scaling the infrastructure are cost and effectiveness. Bandwidth might not be terribly expensive, depending on location, but redundant architectures are expensive to build and maintain. And on top of that, adding bandwidth may not work in every case. Voice and video applications need a minimum level of throughput, but “additional bandwidth alone will not alleviate performance problems such as latency and jitter,” the report says.

The management-focused approach involves performance management software and application traffic management tools. Fault and performance management software alert network managers to device failures. Correlating application performance with network failures helps network managers pinpoint the source of problems.

But these tools remain mostly reactive, in that they alert administrators to a performance issue after it happens. The responsibility to isolate and fix a problem still lies with the network manager, and it can be a manual and time-consuming process.

Traffic management, another option, involves a combination of detailed application discovery and performance monitoring, which can help to establish quality of service (QoS) policies, traffic shaping and compression. QoS policies help network managers prioritize traffic based on its relative importance to the business. But this approach does not guarantee a level of bandwidth for an application, so high-priority applications will still experience performance degradation.

Traffic shaping takes performance management a bit further by guaranteeing bandwidth for certain types of traffic and limiting available bandwidth for others. Several technologies are part of traffic management, Yankee says, such as traffic shaping compression. Traffic shaping can provide a way to prevent congestion within an enterprise network and minimize the impact of rogue traffic on mission-critical applications.