For starters it would to know the conditions under which the test is performed.
If you are using a remote DB with 300ms of lag from SFS2X that is already a significant problem, as every client request will incur in that much latency plus everything else.
That is not the problem. It is expected.
As regards the number of connections being only 2 while the server is being spammed, that doesn't sound right.
Where exactly are you getting the number of connections from?
And what is the configuration of the DBManager on the SFS2X side?
This is from runtime zone settings. 80 Active, 20 Idle, GROW, 3000 block.
The response time should be 500ms tops, but I am getting responses that range between 2 - 10 seconds.
That sounds more like a problem with the database itself than anything else.
By the way what database is this exactly? Where is it hosted?
No, it is a problem because there are only "2 ACTIVE" connections. Once the active connections bumps up (highest I saw was 10), everything is normal where the expected response time is actual. ECS DB, where server and db are hosted in different regions.
##Update
So I am able to get my idle connections up by not closing the db connections until all are created. I am now noticing the SAME problem.
FREE connections are not being used and it keeps on waiting for the same active connection causing huge response times!
I have 20 idle db connections before test start, once the test runs, I see
Active : 1-3
Idle: 16+
Again, it takes a few minutes before I see active connections go to 15-20.
What is going on? Why is this happening?