The "disruptor" pattern has some significant problems with tail performance (around pathological OS scheduling) if you aren't running one thread per core the wa

The "disruptor" pattern has some significant problems with tail performance (around pathological OS scheduling) if you aren't running one thread per core the way LMAX designed it.