A question about the interpretation of difference between purepath response time
Hello Compuware staff,
We have setup the a Business Transaction LoginTxn that measures the response time of the Login PurePath. There are two ways to access our web servers. The first way is to access them directly without load balancer (blue curve). While the second way is to access them through a load balancer (red curve). The PurePath response times for the two access methods are shown in the plot included in this email. Here we can see that the response time through load balancer is longer than that with no load balancer. I am wondering whether the time differences can be attributed to the time spend within load balancer.
Thanks a lot.
Troy
Hi Troy
I think your assumption is correct. The time can be attributed to the load-balancer + the network between your load balancer and your webserver.
Andi
Andi,
Thanks a lot. This could be an interesting way to evaluate hardware performance.
Troy
How are you measuring this transaction? Is your PurePath using UEM so that PurePath starts in the browser? If so, you are on the right path. Look at the transaction flow dashlet and click the option to show inter tier times. The delay of the load balancer should be captured in the inter time time value.
If you are only using a java agent on the backend (no UEM), then your transactions are being measured on the backend only (behind the load balancer). So the load balancer delay is not included in the PurePath measurement in this case.
