Finding the real causes of network bottlenecks

App performance depends on more than just infrastructure

Tags: Intergence SystemsNetwork optimisation
  • E-Mail
Finding the real causes of network bottlenecks Dr Steve Turner of Intergence says companies need to understand that a number of factors can affect application performance.
By  Dr Steven Turner Published  August 28, 2012

When application performance degrades, both users and management can be quick to blame the network, but often there are other factors impairing performance. The important thing is to detect these problems as quickly as possible in order to troubleshoot them.

Before any intelligent techniques can be used to optimise application performance, the key factor in identifying how to improve performance is to first understand the application usage profile of the network and the current level of performance.

Performance should be the baseline prior to any infrastructure upgrades. Traffic across the network and application performance should be recorded and analysed to create an understanding of the current performance profile.

It should then be monitored to ensure performance does not degrade over time. Examples of this analysis can include: How much bandwidth does a particular application use? Is the WAN link suffering congestion? What is the latency and jitter across the network for specific applications?

Per application, Service Level Objectives should be defined to track application performance over time and alert managers when performance degrades or be used to protect and prioritise business critical applications.

Some applications are more ‹greedy› with regards to bandwidth consumption than others. For example, a voice session may only consume 84Kbps of bandwidth, yet a file transfer will consume as much as physically possible without any controls in place, and this can impact on the performance of other more ‹sensitive› applications.

Some applications are more sensitive than others; whilst file transfers can cope with high latency / jitter and even packet loss, real time applications such as voice and video cannot, and the user experience can become very poor. To address this, business critical applications should be protected and prioritised from non-critical applications through optimisation techniques such as WAN Governance.

High latency, high bandwidth links can also impact the performance of applications, particularly those which rely on the TCP protocol to acknowledge packet receipt before sending the next piece of data. WAN Acceleration (or optimisation) can be used to minimise the ‹chatty› nature of many TCP applications, where there is a lot of interaction between the client and server.

These include file transfer applications that will aim to effectively ‘fill the pipe’ and send as much data across the WAN as is possible. Further techniques can be used to compress and de-duplicate data so that only unique data is transferred to the remote host, reducing the amount of data needing to be sent and effectively increasing performance.

Organisations that have only employed monitoring through visualisation have, in many cases, been able not only to improve their application performance, but also reduce their spend on network upgrades. This is because they can restrict or remove non-business activity either through user education or policy enforcement.

Organisations that move to using optimisation techniques, such as WAN Governance and Acceleration, can achieve significant improvements in the performance of business critical applications, whilst reducing bandwidth consumption and increasing throughput. Optimisation can be used to delay or negate the need to upgrade the network, often producing both a rapid return on investment and improvement in end user experience.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code