DotScale 2015 : Leader Election with Cassandra

In every distributed architecture we need, from time to time, to elect a leader amongst servers.

In this talk I will show how to implement a classic leader election with Cassandra. No need to install Zookeeper anymore!

That implementation will use Paxos a distributed algorithm.

Transcript

1. @mNantern Leader Election with Cassandra
2. @mNantern cqlsh:test> INSERT INTO lock (lockname) values ('mon_lock') IF NOT EXISTS; [applied] ----------- True [applied] | lockname -----------+---------- False | mon_lock CREATE TABLE lock ( lockname text PRIMARY KEY ) with default_time_to_live = 180
3. @mNantern Is this really working ?