[LRUG] API automation / load testing

Sidu Ponnappa ckponnappa at gmail.com
Mon Jul 2 04:01:41 PDT 2012


I've had good experiences with Grinder[1] but that was six or seven
years ago. Worth taking a look at, though.

Best,
Sidu.
http://c42.in
http://sidu.in

[1] http://grinder.sourceforge.net/g3/whats-new.html


On 2 July 2012 16:03, Ed James (Alt) <ed.james.spam at gmail.com> wrote:
> Thanks Alex. I've been using jMeter for a while now and for the most part
> it's fine. I was wondering if there were any more Ruby/Rails oriented tools
> that could be incorporated into our project, but it looks like for now I'll
> be sticking with jMeter for development and probably blitz.io for
> production.
>
> Tourbus looks very cool though, so I'll take a closer look at that too.
>
> Thanks.
>
> --
> Ed James
> Sent with Sparrow
>
> On Monday, 2 July 2012 at 09:53, Alex Reis wrote:
>
> Might be overkill for your needs, but I tend to use Apache JMeter a
> lot, since that lets you record (and edit) user sessions, then play
> them back with a ramping up number of threads, different scenarios and
> with some programming capabilities (loops, ifs, variables). It can
> also be used to coordinate a test launching in multiple hosts so you
> don't get to exaust the capacity on your machine before the server
> feels it like others said.
>
> There's also the Tourbus library for what it's worth -
> https://github.com/dbrady/tourbus/ - if you have good capybara-ish
> tests, you can turn them into realistic load test scenarios easily.
>
> Cheers,
>
> Alex
>
> On Mon, Jul 2, 2012 at 9:53 AM, Ed James (Alt) <ed.james.spam at gmail.com>
> wrote:
>
> Thanks for the feedback all - really helpful.
>
> --
> Ed James
> Sent with Sparrow
>
> On Friday, 29 June 2012 at 19:29, John Arundel wrote:
>
> On 29 Jun 2012, at 17:53, Stephen Bartholomew wrote:
>
>
> What about using ab (apache benchmark)?
>
>
> +1
>
> I've also used httperf, and autobench (a wrapper allowing you to co-ordinate
> httperf attacks from a whole bunch of machines). I usually find that I'll
> exhaust the outbound bandwidth of my load-generator machine well before I
> hit capacity on the server I'm testing, so a distributed autobench setup is
> useful in this situation.
>
> You can generate realistic user sessions by taking snippets of your web logs
> and using httperf to replay them (for example, logging in as a test user,
> viewing some product pages, adding some items to your shopping cart, going
> to the checkout - stuff that exercises the whole stack and generates
> database writes). I find this often gives very different (and sometimes
> alarming) results compared to simply fetching the home page a million times.
>
> Regards,
> John
> --
> Bitfield Consulting: we make software that makes things work
> http://bitfieldconsulting.com/
>
> _______________________________________________
> Chat mailing list
> Chat at lists.lrug.org
> http://lists.lrug.org/listinfo.cgi/chat-lrug.org
>
>
>
> _______________________________________________
> Chat mailing list
> Chat at lists.lrug.org
> http://lists.lrug.org/listinfo.cgi/chat-lrug.org
>
> _______________________________________________
> Chat mailing list
> Chat at lists.lrug.org
> http://lists.lrug.org/listinfo.cgi/chat-lrug.org
>
>
>
> _______________________________________________
> Chat mailing list
> Chat at lists.lrug.org
> http://lists.lrug.org/listinfo.cgi/chat-lrug.org
>



More information about the Chat mailing list