While executing a performance test of 150 users using SilkMeter for Kony Web Application, we came across a strange behavior where services are held on wait state on the doFilter method.
The application is running on IBM WebSphere 8.5.10 and web thread is configured as min 50 and max 800. The issue is when we increase the number of users, the wait() method is also increasing in time. Below is a screen captured from Dynatrace. [![Dynatrace report][1]][1]
We tried many methods but couldn't get rid of this wait method. Any suggestions?
This is the expected behavior of Kony which is tested in 7.3.0 version. This happens when the Transactions per hour increases. When there are many server hits, due to kdw, we can see a wait time. If this increases beyond, 5 seconds, then it's better to reduce the Transactions and think time of silk meter. Kony supports 35 requests per sec for 1 CPU and 8 GB memory. Exceeding this would increase wait time as shown in the figure.