This has been submitted to HN a few times before (search: lmax.html)<p>There was a pretty good discussion on this particular thread: <a href="https://news.ycombinator.com/item?id=3173993" rel="nofollow">https://news.ycombinator.com/item?id=3173993</a>
I would like to see a Call Me Maybe-style test of the automatic failover from one disruptor to the next. At 6 million TPS I don't see how they could stay exactly in sync. Is the output queue allowed to emit duplicates?