We are actually comparing our IIS logs requests' time-taken field with the spikes recorded on our application in PRTG. We compiled a couple of days' data from PRTG in csv format and observed the spikes and its respective loading time. Which we compared it with the request log in our application's IIS log file. We found major difference in both the values, not just for one or two entries but for major entries. The time recorded in our IIS logs is much lower than what is recorded in the PRTG log.

For e.g.

According to PRTG, below are the date-time and the load time of the request.

Date Time LoadTime

12/9/2012 12:42:00 AM - 12:42:30 AM 807 msec

12/9/2012 12:42:30 AM - 12:43:00 AM 858 msec

12/9/2012 12:43:00 AM - 12:43:30 AM 811 msec

Whereas our IIS logs record the following information for the above mentioned time intervals.

2012-12-09 00:42:19 W3SVC2 10.63.88.111 GET / - 80 - 202.160.175.2 Mozilla/5.0+(compatible;+PRTG+Network+Monitor+(www.paessler.com);+Windows) - xyz.com 200 0 0 28038 204 671 Time Taken - 671 milliseconds

2012-12-09 00:42:50 W3SVC2 10.63.88.111 GET / - 80 - 202.160.175.2 Mozilla/5.0+(compatible;+PRTG+Network+Monitor+(www.paessler.com);+Windows) - xyz.com 200 0 0 28039 204 703 Time Taken - 703 milliseconds

2012-12-09 00:43:20 W3SVC2 10.63.88.111 GET / - 80 - 202.160.175.2 Mozilla/5.0+(compatible;+PRTG+Network+Monitor+(www.paessler.com);+Windows) - xyz.com 200 0 0 28038 204 671 Time Taken - 671 milliseconds

Would you please give us your expert view as to why we observe such differences in time taken for the request?

Is it like that PRTG records time taken for full loading of the page i.e. time taken till the last byte is received, so it shows more time consumed than what the IIS logs show?

Any help is duly welcomed.

Thanks in advance,

Dhaval

Article Comments

Hello,

which exact HTTP-Sensor are you then using in PRTG?

Please bear in mind that the difference are merely 150 milliseconds. That could easily be a slightly different way of measuring times, or using different performance timers to actually measure the time.

best regards.


Dec, 2012 - Permalink

Please keep in mind that IIS can only measure the time it took to generate the page and hand it over to the IP stack. IIS does not know how long it took for those bytes to travel to the client system.

PRTG is on the other end of the pipe. It measures the total time for all the following steps:

  • resolve the server name using a DNS query
  • open the TCP connection to the web server (your IIS)
  • send the http request header
  • wait for a reply (that's the time you see in the IIS logs)
  • receive the http reply header
  • receive the http payload/web page
  • close the TCP connection

Depending on the amount and speed of the infrastructure between the PRTG probe system and your web server 150ms extra can be just about normal.

If you use the full page sensor then the probe actually requests the page with all it elements (css, javascript, images etc.) and also waits until the browser has fully rendered the page. So that's a completely different thing by itself.


Dec, 2012 - Permalink