HAProxy manages internal and origin load balancing using KeepaliveD. Two small servers host the entire site, never moving about 15% load even during the largest load spikes.

Akamai sits at the front of the site, all assets and pages are cached for a minimum of 60 seconds. Requests to the edge peak at over 100k/s.

Memcache caches database results and articles, reducing overall DB load and allowing seamless DB maintenance during quiet periods.

Full text search is provided by a SOLR cluster. This is done on Master/Slave replication with Varnish as a cache.

Apache splits static traffic from application traffic, as well as providing a selection of tools to assist in running of the site (rewrites, logging etc).

Tomcat provides a fantastic application container that runs the majority of services of the site.

Java powers the Escenic CMS used in the stack.

MySQL is the primary datastore, this is setup in a Master/Slave configuration, with offsite replicas.

Varnish sits as a secondary cache layer behind Akamai. Two servers operate in a Primary/Secondary configuration with failover managed by HAProxy. Requests peak at around 10k/s.

vSphere provides a central management interface for the entire stack. In addition to application level redundancy, VM level redundancy is offered through HA and DRS.