Comment by sesm

When I worked with Erlang outside of Ericsson, I’ve heard the following from our tech lead: OTP built-in distributed process and failover mechanisms assume much lower latencies than usually exist between datacenter VMs and the underlying protocols consume a lot of bandwidth. I didn’t dig deep enough to verify this, but I can confirm that typical Ericsson hardware is several boards connected with high bandwidth ‘backplane’ and OTP failover was designed for this environment. VMs communicating through LAN is a different environment, and VMs living in different data centers are very far from this. I wonder if anyone else tried using OTP outside of Ericsson and what is their experience.