Whether you’re developing applications or delivering applications, understanding how they will perform over the real world WAN is critical to your success. This has been the cornerstone of Shunra’s philosophy for over a decade and was recently discussed in an article by NetQoS.
In the article, Patric Ancipink says that application response time is the most important metric for network performance management. He further explains that the application response time is mostly affected by the latency of your connection and the number of connections the application makes (application turns). He is right. There are things that you can do to improve, or reduce your latency between two locations, but there will be a point where you simply can not improve further. Likewise with the number of connections; there are many techniques you can try to reduce the number of connections and the number of interdependencies on the connections you use, but you will eventually hit a point of diminishing returns.
The logical next step is to start trying to test for these effects and see how your application performs already. Shunra has a great tool called VE Analyzer that helps you do this. It works with our appliances and desktop WAN emulation solutions, so it’ll fit into your existing testing or offer an easy entry into application delivery testing. I don’t want this blog to sound too “salesy” so I’ll leave the deployment information at that. What VE Analyzer will do is gather the packet data from your test and then tell you things like where your packets are spending their time, how many application turns each transaction is taking and how long each turn takes, where any network bottlenecks exist, etc…
Quickly generating these metrics enables you to understand the most important things: how sensitive your application is to latency, and how many connections each transaction is using. You can then either make changes to your application and re-test or try what-if scenarios by changing the emulated network. Either way you choose to address the problem will give you valuable insights with ease and efficiency of time that you can not find anywhere else.
How are you currently testing application delivery? Do you consider latency, number of connections, and application response times? I’d love to hear about your projects or thoughts on this post.
Leave a Reply