Timeline for Load testing : how to generate per second requests?
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 8, 2012 at 16:46 | vote | accept | King | ||
| Jan 30, 2012 at 0:04 | history | bounty awarded | King | ||
| Jan 30, 2012 at 0:04 | vote | accept | King | ||
| Jan 30, 2012 at 13:55 | |||||
| Jan 20, 2012 at 22:19 | comment | added | King | Right on the nail. This is exactly my problem in creating such a set up for load testing. The guarantee to so and so number of requests is puzzling me. The measurement and assert of per second will only slow down the number. I'm actually looking for something to actually beat me to the wall saying that it actually is possible. | |
| Jan 20, 2012 at 18:32 | comment | added | Christopher Currens | @King You can't guarantee it. You can get closer to the number of requests you want if you use constructs that are non-blocking, like above to improve the speed of the requests so you can actually hit that number. You can do checks to see how many have been run in a period of time and try to throttle it, but if you aren't hitting that number anyway, that's a problem. The problem with checking and comparing number of requests to a span of time is that the check costs CPU time, and can wind up limiting how many requests you can send in the first place, which might not be the requested N. | |
| Jan 20, 2012 at 17:59 | comment | added | King | I have no problem in actually sending them over ICE. The problem is what defines the implementation that would actually create N requests and something that can be said true to that number,N. | |
| Jan 20, 2012 at 17:43 | history | edited | Christopher Currens | CC BY-SA 3.0 | deleted 1169 characters in body |
| Jan 20, 2012 at 17:13 | comment | added | Christopher Currens | @King - I guess I wasn't as clear as I thought I was. The last 3-4 paragraphs I thought would help you. I had assumed you were already using a loop of sorts already. If you were doing that, the problem is that your socket sends/receives are blocking, and thus slowing your requests down. Maybe I'll find some time to post some example pseudo code. | |
| Jan 20, 2012 at 9:29 | comment | added | King | I was forced to reduce my questions because they wanted to vote it to close. Now it seems, its valid here to have them. @ChristopherCurrens +1 good point for difference with threadpool to tasks. That widened my understanding. But I'm still stuck how generating some N requests per second is really possible ? What exactly is the best way to do that ? | |
| Jan 20, 2012 at 8:42 | comment | added | Christopher Currens | @svick well the ops original post had 4 questions originally, and they asked questions about the performance of parallel vs tasks, then it was edited out, and now theyre back in. So, much of what you read was a result of that. Ultimately, though his question does have to do with performance, as he has the general idea correct, but is apparently lacking in his implementation. I believe my pointed answers at the end answer the question he didn't edit out. | |
| Jan 20, 2012 at 7:57 | comment | added | svick | Why are you talking about performance so much? That doesn't seem to be what the OP wants. | |
| Jan 20, 2012 at 5:14 | history | migrated | from stackoverflow.com (revisions) | ||
| Jan 19, 2012 at 22:48 | history | answered | Christopher Currens | CC BY-SA 3.0 |