Re: [OpenSTA-users] Network delays
Brought to you by:
dansut
Menu
▾
▴
Re: [OpenSTA-users] Network delays
|
From: nirav s. <ni...@ho...> - 2003-04-29 15:13:23
|
<html><div style='background-color:'><DIV> <P>Thanks Dan,</P> <P>I will work with SNMP collectors, and see what I find.</P> <P>Nirav<BR><BR></P></DIV> <DIV></DIV> <DIV></DIV>>From: Daniel Sutcliffe <DANSUT@OPENSTA.ORG> <DIV></DIV>>Reply-To: ope...@li... <DIV></DIV>>To: ope...@li... <DIV></DIV>>Subject: Re: [OpenSTA-users] Network delays <DIV></DIV>>Date: Mon, 28 Apr 2003 23:48:52 +0000 <DIV></DIV>> <DIV></DIV>>nirav shah wrote: <DIV></DIV>> > I am running the load tests but my results are affected by the <DIV></DIV>> > variations in the network delay. <DIV></DIV>> <DIV></DIV>>I'm guessing you mean network delays that are outside your control, <DIV></DIV>>not network delays caused by your testing. It sounds like you are <DIV></DIV>>doing load test over an unpredictable network, say the Internet. <DIV></DIV>>This is normally going to be less than ideal, there are just too <DIV></DIV>>many unknowns involved. <DIV></DIV>> <DIV></DIV>> > Is there a way I can learn how to get an estimate of network <DIV></DIV>> > delay, and make my results independent of network delays ? <DIV></DIV>> <DIV></DIV>>If the above info I provided is a true estimation of the situation <DIV></DIV>>you have, then what you want to cancel out are just the unpredictable <DIV></DIV>>network delays. You don't want to cancel out delays that you are <DIV></DIV>>actually causing yourself by doing the load testing, these are a <DIV></DIV>>valid part of your results. This is going to be v.difficult! <DIV></DIV>> <DIV></DIV>>What will help you, is collecting lots of data using the NT <DIV></DIV>>Performance and SNMP collectors from stations throughout your test <DIV></DIV>>network. Perhaps create a really simple script which loads a small <DIV></DIV>>static page off your Web server and run this repeatedly as part of <DIV></DIV>>your load test. Also follow all of Bernie's suggestions. The above <DIV></DIV>>data will only give you an idea of the amount of "background noise" <DIV></DIV>>going on during your load test by back to back comparison of <DIV></DIV>>equivalent test runs. I really don't know how you can get a "delay <DIV></DIV>>number" to subtract from your desired results without knowing much <DIV></DIV>>more about your environment and tests. <DIV></DIV>> <DIV></DIV>>Maybe someone else has a formula or method for working out the <DIV></DIV>>unwanted delays without including the wanted (load testing created <DIV></DIV>>delays). The ideal method is using a controlled testing environment <DIV></DIV>>with a known quantity (zero) of unwanted delays. <DIV></DIV>> <DIV></DIV>>Please let us know how you get on and what helps you clean up your <DIV></DIV>>results. <DIV></DIV>> <DIV></DIV>>Cheers <DIV></DIV>>/dan <DIV></DIV>>-- <DIV></DIV>>Daniel Sutcliffe <DANSUT@OPENSTA.ORG> <DIV></DIV>> OpenSTA part-time Caretaker <DIV></DIV>> Visit the OpenSTA Community Site - http://portal.opensta.org/ <DIV></DIV>> <DIV></DIV>> <DIV></DIV>>------------------------------------------------------- <DIV></DIV>>This sf.net email is sponsored by:ThinkGeek <DIV></DIV>>Welcome to geek heaven. <DIV></DIV>>http://thinkgeek.com/sf <DIV></DIV>>-- <DIV></DIV>>OpenSTA-users mailing list Ope...@li... <DIV></DIV>>https://lists.sourceforge.net/lists/listinfo/opensta-users <DIV></DIV></div><br clear=all><hr>Find a partner. For life. <a href="http://g.msn.com/8HMAENIN/2755">Meet@Shaadi.com</a> </html> |