Hystix

[Roman Leventov]

While reading this blog post on sync vs async: http://techblog.netflix.com/2016/09/zuul-2-netflix-journey-to-asynchronous.html (highly recommended btw) I stumbled upon Hystrix: https://github.com/Netflix/Hystrix, a harness for coping with issues of sync services/distributed systems.

There is no "magic": https://github.com/Netflix/Hystrix/wiki#how, https://github.com/Netflix/Hystrix/wiki/How-it-Works. Basically: thread pool per remote service, timeouts, latency/error statistics.

In Druid we already have some of those ideas applied in various forms, however (seems) a big added value of Hystix is ready-to-use dashboard for cluster monitoring: https://github.com/Netflix/Hystrix/wiki/Dashboard

IMO it makes sense to integrate Hystrix into Druid for having this Dashboard on hand.