Scaling Concurrently

I’ll discuss many of the issues involved when it requires more than one, ten, or hundreds of machines to service web requests and the battle-tested solutions to allow those systems to communicate together rapidly, redundantly, and securely.

Topics will include:

• Current technologies for interconnecting systems (Thrift, Protobuf, Memcache, Zookeeper …and more!)

• Understanding timeouts in the request/response pipeline

• Redundancy and service discovery

• Abuse mitigation at large scale

• Load balancer concurrency and connection rates

• Dealing with an avalanche of logging data

• Low level TCP debugging to understand system performance

• Monitoring and reporting of cluster performance