We have a .net application that calls a web service (that we have written) via https. The web service is installed on two machines (API01 and API02) and these are load balanced over SSL on 443
We have a test harness which calls the API which is run currently from a single machine.
If we amend the DNS setting so that the URL points at one of the API machines the API calls succeed and an response code of 200 is returned.
When the DNS setting is amended to use the load balanced VIP between the two API machines we have noticed that some of the response times take a long time, this has caused our automated regression test pack to fail. This isn’t consistent but the values range from a few milliseconds (which we expect) to 92 seconds.
When we look at the logs on IIS for the actual requests IIS implies that the request has been dealt with in normally less than 30 milliseconds.
Any ideas to improve this?
Netscaler 7500 x2 HA. NS10.5 61.11.nc