Sphene Community Tools

I am curious about the load this website has on the server. It seems to take me several seconds to make a post. Thats a fairly big consideration as to whether or not I'll end up using this (or django in general). Can anyone provide any details?

the load is generally 0.8-1.0 .. this server is not optimized for performance.. it's just a playground for me :) it runs apache2, tomcat, my own java web server, jabber server, various gateways, mail server, an insane number of munin scripts which run every 5 minutes, etc.etc.

there are of course many things which can be optimized in SCT (especially the list of threads in a category .. currently has a very high number of SQL statements which are basically useless and could be easily optimized (by using select_related)) .. but i would suggest to try it out yourself before using my server to get performance benchmarks :)

i haven't yet optimized SCT for performance (except a few obvious places where i added caches)... so if you have any problems with it let me know :)


P.S.: it seems your POST needed 817ms + 1492ms for the next GET (including sending, not just processing) .. so your impression wasn't that wrong ..

--- Last Edited by Herbert Poul at 2008-06-09 08:01:00 ---

Instant response! Awesome. ;)

Yeah, I am interested in starting a community site and was planning on providing various wiki's, forums, IM, etc.. Django experience is limited but I like what I see and I know python pretty well. My only concern was the speed of ORM's as opposed to traditional bare-bones SQL I'm used to dealing with (LAMP mostly). I'll give it an install and kick around the tyres when I get home tonight. ;)

Thanks for your comment!

FYI, as I implement the various features of my site I intend on doing some serious complexity and performance testing. I'll be sharing as I go.

My general measures are:

• request response time under: no load; 50, 100, 500, 1000 requests per second
  
• System load under 50, 100, 500, 1000 requests per second
  
• Database load under 50, 100, 500, 1000 requests per second
  

I also measure complexity and performance at the method levels which deal with:

• cyclomatic complexity
  
• Fan-in / Fan-out complexity
  
• middle ware database usage (number of connections / queries)