Application Response Time: Buyer Beware!

Veena Vadgama, marketing director for Network Instruments, covers the subject of application response time and why vendors’ claims of accurate metrics for any application may not be true.

Does your monitoring device claim to measure application response time or ART? This is great, because measuring ART is the first critical piece you need to fully understand network performance and overall health. But we bet you think your tool monitors user response, right? Well, in most cases, you’re wrong.

Make note of this important fact: if your product only measures Layer 4, you will not be able to see any server problem issues, which means you completely miss how your users are affected by applications.

Using SQL as an example, true application response time measures how long it took the server to deliver every last bit of data asked for a client’s SELECT request, down to the last data frame.

Some tools merely measure TCP handshake completion time or TCP ACK response time and call this ART. While you will get TCP performance, that is all you will get – no user response information.

Calculating true application response time requires that the analysis tool have some hard-coded knowledge of the monitored application and how its clients and servers communicate. For this reason, it is typically available for common applications such as SQL, HTTP, FTP, etc.

CAUTION: If an analysis tool claims to measure application response times for any and all custom applications, what it really shows is probably some form of TCP response time.

We’re proud to promote the fact that Network Instruments’ Observer Expert provides true application response time and analysis by using built-in intelligence to appropriately understand how specific applications, including SQL, Citrix, MS Exchange, HTTP, and FTP, interface with clients and servers.

See a list of supported applications on our Observer Expert page.