Scalability and Performance Options

NodeOne - Tomas Persson
Talking about how Drupal could move towards the cloud and what we need to think about when we scale a Drupal site for 100.000 users.

Voxel - Sam Machiz, VP of Sales Engineering
Brands such as FastCompany, New York Observer, Human Rights Watch and the National Review all count on Voxel to deliver a high performance Drupal experience for their users. We will look at the technical architecture required to deliver this scale of traffic.

Adyax - Maxime Topolov, Co-founder of Adyax
Looking at the usage of MongoDB, Varnish, Memcached and specific modules optimisations on www.freerice.com, a 20 millions page views per day Drupal site.

Axis12 - Dave Stuart, CTO of Axis12
A large UK Government website receiving 30 million pages views a month was struggling under the load and the Axis12 team helped to improve performance. Through the use of Memcache, Varnish, S3 and ESI they were able to significantly reduce server load.
Issues:
- Multi surfacing of content in different sections of the site meant a straight page cache wouldn't work, because content had context
- The site was serving over 1TB of downloads a month which was costing a lot in bandwidth and performance
- Bottlenecks and lock ups were occurring randomly and frequently on the site

Microsoft - Mark Brown, Sr Product Manager
The ability to go and build a scale out Drupal site on Windows and IIS is something that only became possible just recently. We’ve built out a lab environment where we run this and have just created documentation that we are planning to add to the Installation docs on d.o here in the next few weeks.

Intended audience

Questions answered by this session

How Drupal is different from any other solution in terms of performance?

What are the major differences in addressing performance issues between anonymous and connected traffic?

How do we scale Drupal for logged in users?

How do we handle caching for logged in users?

What kind of technical architecture approaches help with regards to delivering guaranteed QoS in the face of high-volume but spiky traffic?