www.drupal.org traffic & resources
tesliana - May 26, 2007 - 21:40
Folks,
In order to be able to plan a little better for a Drupal powered community site, it would be valuable to me to know how many daily or monthly visitors www.drupal.org gets and the specs of the hardware and bandwidth that it's hosted on. Is there anybody out there who is willing to share their own numbers even if without disclosing the web sites involved. The more detailed the info the better.
Thanks a lot.

not sure about d.o but..
You might want to take a look at the following thread:
http://drupal.org/node/116578
It goes into quite a bit of detail for TeamSugar, a community based site built in Drupal. It sounds like they were doing about 1.1m pageviews/month on TeamSugar alone, and 21m on their entire network (which is all Drupal).
subscribing
subscribing
I'll chime in, this is quite
I'll chime in, this is quite well dicussed but may as well have a go...
For one set of sites we run...
Server: Dual Intel Xeon 3.0GHz server with 4GB DDR ECC Reg. RAM, 3 x 73GB
Hardware RAID Protected SCSI Hard Drives with Battery Backup. 500GB
Monthly Data Transfer. Big fat pipe to internet, located in Tampa, Florida.
Runs 2x Drupal 4.7 sites and 1x non drupal site.
Drupal site 1 - around 450 000 pv pm (uses 60% + of resources)
Drupal site 2 - around 20 00 pv pm (minimal resources used)
Non drupal - around 550 000 pv pm (uses around 35%)
Yes you are reading correct - the Drupal 4.7 site uses nearly double the server resources as the non drupal CMS for less page views... and (read on) this is with blockcache used on nearly every block, very few contrib modules & the Boost File based cache system! (without boost we would be fecked, no ifs - no buts, I dont care what anyone says, our hard worn experiments show us the database sobs, weeps and then commits suicide if we turn off boost and rely on drupal 4.7 cache system).
All 3 sites run about the same ratio of logged in users to anon (both serve cached pages to anon, uncached to logged in users). It about 95% anon traffic, so they are not really community sites in the true sense of the word, but of course you require far more server resources to run Drupal as a large community site.
We run the Boost file cache system set to expire 30 mins. To start with we only cached nodes but now we cache everything. Using the standard drupal cache is not an option. We're strongly considering moving to D5 and SQUID or other reverse proxy cache. We have grown to this size in only 8 months and we know there is considerable growth to be had.
We've had a lot of problems with bots hammering the server, so prodigious use of robots txt and .htaccess is essential.
Admittedly the code is not well optimised for performance, the rather limited budget hasnt allowed for much there.
I'd really like to hear from others in the same position also. Frankly our biggest issue is the sheer number of queries drupal requires just to genearte a node - we dont use that many modules and very few contrib modules at that. We do use tagadelic (1x cloud set to refresh every 2 hours with blockcache).
The big hogs appear to be Pathauto & Views (probably mainly the way we have used Views). We cant live without pathauto so that is out next taget, we really need to look hard at path aliasing and performance.