Web Service gets unavailable after several concurrent calls

Posted by Roman on Server Fault See other posts from Server Fault or by Roman
Published on 2012-04-07T15:29:54Z Indexed on 2012/04/07 17:34 UTC
Read the original article Hit count: 352

Filed under:

We are testing GoDaddy Virtual Data Center and came to a very strange issue when our web site gets unavailable. GoDaddy Support keeps saying the issue is in our web server settings, but looking at the result of our tests I doubt it.

TEST ENVIRONMENT

  • Virtual DataCenter with Windows hosted at GoDaddy.com. All servers have Windows Server 2008 R2 Datacenter, IIS 7.
  • Server One with IP address 10.1.0.4
  • Server Two with IP address 10.1.0.3
  • Both servers are in private network not visible from outside.
  • Port Forward with IP address 50.62.13.174. Port Forward is assigned to Server One

TEST DESCRIPTION

  • JMeter is used as a Client App to simulate 30 concurrent users sending 100 SOAP requests each. Interval between requests is 1 second.
  • Http link used for testing: http://50.62.13.174/v2/webservices.asmx

TEST ONE

  • Test is run from a computer in our office.
  • After JMeter starts running test, almost immediately, the link above becomes unavailable in a browser.
  • After test completion, the link is not available in a browser for about 5 more minutes.
  • Remote Desktop is working well, so we can connect to Server One remotely.
  • After about 5 minutes since test completion, the link becomes available in a browser again.

TEST TWO

  • Test is run from Server Two (that is part of our virtual data center).
  • Test works very well, no visible delays in processing.
  • The link is available in a browser all the time.

TEST THREE

  • Test is run from Server One using localhost.
  • The result is the same as in TEST TWO - no issues.

TEST FOUR

We repeated TEST ONE from other computers that we have located in different countries, all with the same result as TEST ONE.

CONCLUSION

As the test works well from Server Two, but does not work from outside our virtual data center, we feel there are issues with the network or its capacity. The whole behaviour looks like out requests from outside get stuck somewhere before reaching our virtual data center.

Has anybody had similar issues in the past? Are there chances that something is wrong with our server settings?

© Server Fault or respective owner

Related posts about Performance