Multi cloud swarm latency issue #132
Replies: 2 comments
-
There are multiple problems that exist with Swarm, Kubernetes, and other orchestrators when they span a high latency link. In general, it's best to keep node-to-node communications below 10ms (one cloud region).
If using Swarm, I recommend using Portainer to manage multiple clusters from a single UI :) |
Beta Was this translation helpful? Give feedback.
-
I was going to respond before but I wanted to check out portainer before I did but I haven't had time to do that yet... soo.. |
Beta Was this translation helpful? Give feedback.
-
A lot of people say that swarm shouldn't be run on multiple clouds because you get problems with high latency. I've never really seen it explained why that is.
Is it because of the raft protocol? If so what if I just put my worker nodes on other clouds but I keep all of my manager nodes in one cloud's low latency network (or just have one manager node)? Would that make it ok to run swarm across multiple clouds? I know there are high availability issues. Like if there is a network partition between my masters and my workers then things won't work.
Beta Was this translation helpful? Give feedback.
All reactions