Split up the Response Time into Network and Server Times
We are doing Performance testing of a web-application.
We are getting the total response time from Load Runner
We want the break up of this time as Network Time + Server Time = Response Time. We know the response time.
1) This Network time would be: Insertion Delay+ Queuing Delay+ Distance Delay
Insertion Delay would be: To know the Page size (So, if the page size is 64K over a 128 Kbps line, the Insertion delay time
Would be 64*8/128 = 4 secs.
Queuing Time would be the factor I want more info from u guys: (From my understanding and googling it needs Packet Size to be one of the Input Parameters, how can I find out this info?
Distance Delay would be the factor which can be neglected for the Kind of network we are doing Load testing.
(We are doing it in an isolated network and all servers are located within reach, I believe it makes sense to calculate Distance Delay if the
Servers are located far-off in a WAN scenario and thus Distance from the client to server comes into picture; so we are neglecting this)
2) For the second part of my question, is there a way in which we can calculate Server Times also (given the page size or is it that the
Server Time = Total Response Time – Network Time
3) Also, does Time to first buffer make sense in anyway in the above calculations we want. (i.e. for the Server Response Times)
Please provide inputs to this
Thanks and Regards,
Re: Split up the Response Time into Network and Server Times
There are a couple of answers to lead you down that path (not I'm not a LoadRunner expert)
1) I do not believe that LR has the ability to do this (within a reasonable level of effort) on it's own.
2) You can have your developers/admins add time stamps to logs, parse the logs and extract the info you are looking for.
3) You can get a product like PerformaSure to use with LoadRunner to suppliment the collected data. (Borland has a competing tool to PerformaSure - the key is to find one that supports your particular environment).
I am a strong supporter of 3), though I usually end up doing 2).
I don't think any of those options answer all of your questions, but I think they will put you much closer.