[HeyImAlex]

I feel like it's an endless battle, trying to move away from a SPOF; even in your rabbitMQ cluster your load balancer is a SPOF. Who watches the watchmen? At a certain point you just have to accept that you're not gaining a whole lot by adding more complexity. Process supervision over a network (a la something like fleet or flynn or mesos) is a pretty neat solution. I've been playing around with this stuff the past week and I'm having a lot of trouble coming up with the perfect setup haha

[vidarh]

A load balancer need (should) not be SPOF: You run two (or more) of them, with IP takeover (e.g. via keepalived or similar, that continuously vote on the master), or you make the clients try to connect to two or more IP's. Or both.

It's really quite trivial to ensure your load balancer is not SPOF if you're ok with clients having to reconnect on failure, and only slightly more hassle if you want to allow connections to survive one of your load balancer boxes failing.