3

I am a newbie here, please let me know if I'm using wrong webmaster terms.

I am currently setting up a VPS for a multi site WordPress. The VPS uses Debian 6 LNMP setup and the DNS is being taken care by another service. Currently the VPS is running non multi site WordPress with -+ 83 MB RAM out of 128MB. As far as I know the performance is relative to the number of request, not the number of sites in the multi site setup.

The question

How do I calculate maximum number of request in with that setup? If the information is not enough, what other factor do I need to know?

Thank you in advance.

Improve this question

3 Answers 3

Reset to default
3

You could use a stress-tester like this one: http://loadimpact.com/ which has a free option, but limits it to 50 users. Paid-for (or registered users) can get increased stress levels.

I like this service specifically because it generates very useful graphs, and a detailed report containing page load errors, etc.

There are many other services like this one, each with pros and cons.

Improve this answer
3
  • I just tried it. But I'm not sure how to read it. Here is the result. Can you point me how to read this? If im not mistaken, there is no problem to handle 50 active concurrent client?
    – ifdion
    Commented Sep 12, 2012 at 4:43
  • @ifdion - Even at 50 users, I see no signs of increased load (higher ping and load times). This suggests you have a LOT of room for growth. (Consider paying for a test with more clients).
    – ionFish
    Commented Sep 12, 2012 at 19:06
  • This sounds like good news too me. Thank you @ionFish
    – ifdion
    Commented Sep 12, 2012 at 22:27
0

No point in "calculating" when you can just test it, as ionFish suggests—especially as the number of concurrent requests a server can handle depends on the application being hosted (as well as the web server, DB server and their configurations). The same VPS can server 6000 requests per second with static pages or just 1 request per second if it's a very processor/memory-intensive page.

In addition to loadimpact, you can also use load/stress testing and webserver benchmarking tools like:

  • mysqlslap
  • jmeter
  • deluge
  • grinder
  • httperf
  • ab
  • siege
  • valgrind
  • fetch
  • autobench

BTW, all of the above are open source, and more are listed on this page.

Improve this answer
1
  • Thank you for this answer. I believe this means I should just optimize my WordPress caching setup and stop worrying about the number of request.
    – ifdion
    Commented Sep 12, 2012 at 5:25
-1

I wouldn't worry so much about RAM when it comes to a VPS, you need to think more CPU. A VPS with that little RAM likely has almost no CPU. You'd be better off just getting a reseller/shared account (unless your activity prohibits a shared environment). Users are going to be experiencing severe loading issues with that little amount of resources available.

Improve this answer
2
  • The VPS spec says 'Intel Xeon Harpertown Dual Quad Core (8 core total)'. Is this the one I should worry?
    – ifdion
    Commented Sep 12, 2012 at 4:45
  • 1
    It's actually the exact opposite. A VPS isn't a real server. It's just a virtual allocation of resources on a cluster of servers. A cluster of 33 AMD K7s still has as much processing power as a dual quad-core Xeon. And if that K7 cluster is being used by only 20 users while the dual Xeon is host to 50 users, then you're still better off on the K7 cluster. What matters is how much RAM you're guaranteed and also how much CPU time you're allotted. The underlying hardware doesn't matter, and will probably change over time with most hosts.
    – Lèse majesté
    Commented Sep 12, 2012 at 5:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.