High registered-traffic - what setup is best for server, database and software?

Events happening in the community are now at Drupal community events on www.drupal.org.
flollo's picture

Hello High Performance-Group,

we are about to move our client-sites from an only-static-CMS to Drupal and worry how our high registered traffic will do and what would be best setup for Drupal, caching, database- and servercluster-setup. Anybody has ideas and/or experience?

Some figures:

Traffic:
PageImpression-peak per hour: 150,000 (42 per sec.)
PageImpression-peak per hour in closed area/registered user: 75,000 (21 per sec.)
Unique visit-peak per hour: 75,000 (21 per sec.)
Active session-peak (open parallel sessions, timeout 15 min.): 1,500

Content:
Pages: ca. 30,000 (user use a lot of different pages, not just some few favourites)
Teaser: ca. 20,000
Pictures (!): ca. 1,000,000
Picture-Galleries (with ca. 50 thumbnail-pic-teaser on one site): ca. 15,000
Modules: Some standards and some community-functions like non-big forums, setcards and blogs, let's say something every page has today. Galleries are the main thing. We do have some User-Uploads (Text, Pics). We are also running some self-made toplists (e.g. most watched galleries), but that can be cached and just updated every x minutes.

Technic:
CMS: We go for Drupal 6.x
Websites: 3 big ones running on different domain-urls, best to run on one Drupal-installation with possibility to swob content between this three websites. Apart of the "big three" there are some thousands of more Domains for SEO, but they are not big in traffic and cover just non-registered part.
Server: Apache / Linux on IBM-based servercluster is preferred
DB: MySQL 5.x preferred
Loadbalancer: System should work with non-persistent LB-sessions
Edition: There should be one seperate stage-server for editors (not for user out there)
Upload: User are able to upload text and pics (not many user do so, but should work fine within the cluster)
Demand: System must be very quick, stable, reliable and should not fail when some hardware brokes :-)

Anybody out there with experience and/or hints for us? Would be just great - many thanks!!

P.S. Excuse my english.

Comments

Nobody has an idea?

flollo's picture

... still searching for answers, nobody out there has an idea? Many thanks!!!

What are you asking?

boris mann's picture

You're not really asking us anything. We could hand you generic tips -- "don't run too many modules", "tune your database", "install APC/xcache" -- but that's not very specific at all: you could google drupal performance tips and get the same answer.

What is the best setup comes down to your exact needs.

Here are a few random items that might be helpful:

  • check out the deploy module for running your staging server for editors
  • why run your own servers? use EC2/S3, depending on your file serving, the new Amazon CDN Cloudfront might be useful as well

Does it work at all?

flollo's picture

Hi Boris,

many thanks for answer. I'm not asking anything? - I worry I ask to much... ;-)

To be honest the most important question for me: Can Drupal 6.x handle this kind of traffic at all with good response-time? Can I be absolutly sure of that and start the project to move to Drupal?

I searched a lot in the "high performance"-group and via Google, but could not get experience or information about registered traffic (user logged in) in our mesures or above. Caching of more or less static non-registered pages looks pretty clear, but what about high traffic with registered users, especially if it is not forum but most of all pics-galleries?

Would be great to get some hint or experience?

Many thanks for that!!!

No simple answers

boris mann's picture

To be honest the most important question for me: Can Drupal 6.x handle this kind of traffic at all with good response-time? Can I be absolutly sure of that and start the project to move to Drupal?

No one can answer this for you. You can use devel module to generate nodes and users and do load testing on your proposed architecture to get some ideas.

If you look at this group, there are many discussions on different techniques. But, at scale, every site is going to be a little bit different, and a little bit the same. Code tuning, database optimization, sub-page level caching etc. etc. etc.

If you're worried up front, spend some money with some of the existing specialists like Advomatic or Tag 1 Consulting to help you architect and set up. There are some Drupal specific things, but the rest of it will be LAMP scaling.

High performance

Group notifications

This group offers an RSS feed. Or subscribe to these personalized, sitewide feeds: