On 07 Feb 2008 Thu 13:20:47 pgb wrote: > Eren, > > I don't have big experience but I think > there is no way that you can use frameworks and have small amount of > memory and slow processor, especially if your application is big. > > One of my colleagues spend about 4 months writing quite big project in > django. > It was installed on a small box with 512 MB of RAM and Celeron 2.0 and > we asked 40 people to just browse the new intranet website - after few > seconds the server went down. > So we have moved everything to HP Proliant - RAID5, 4GB of RAM, 2x > dual core 3.00GHz and MySQL db also to new machie with the same spec > as web server. > Unfortunately, even on new machine our server was not able to do more > then 5 requests per sec without caching! > What really surprises me is that big django app can use even 90% of > CPUs!
Really thanks for your reply, That's what I saw, in few minutes, my server was down :) > My advice is to us ab (Apache HTTP server benchmarking tool) to see > how many requests per second your site can do. > but the first thing is to buy your servers more RAM and use cache. I'll use this tool to see how strange my server is. It looks really usefull > Just to see how fast caching can be, add the following into your > settings.py: > CACHE_BACKEND = 'locmem:///' > and 'django.middleware.cache.CacheMiddleware' as a first line to > your MIDDLEWARE_CLASSES I have an experience in caching (at least, in how to setting up). I'll just set up a memcached server and install python bindings. I think, caching will reduce resource usage because most of the parts of the site is stored DB and is static. My site contains a lot of documents. > Second thing is to profile your code. > Django by default comes with django.core.handlers.profiler-hotshot > > To see where is the bottle neck of you app > mkdir /var/log/cmsprofile > chown www:www /var/log/cmsprofile > add django.core.handlers.profiler-hotshot > #PythonHandler django.core.handlers.modpython > PythonHandler django.core.handlers.profiler-hotshot > How to read the prof files from hotshot see > http://www.rkblog.rk.edu.pl/w/p/django-profiling-hotshot-and-kcachegrind/ > > You can also use Profiling Middlware from > http://www.djangosnippets.org/snippets/186/ > > Processing requests can be very expensive. > so users_list = get_users(request) > where get_users is large def, will almost always take huge amount of > resources. Thanks. I didn't know this profiling stuff. I hope, I'll find out what parts of code are slow. > Also, especially if your app is big, lines like: > houses = House.objects.all() > groups = Group.objects.all() > is more then likely about 15 req per sec less, even if these are very > small tables. > > kind regards, > Paul Best regards, Eren --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
