This message was deleted.

I remember reading a paper on this. You’d think that for each additional server you’d add, they’d each contribute the same percent to overhead. But this was not the case. From what I remember like adding 1 node to create a 2 node cluster was adding like 5% overhead. But if you have 4 server nodes and wanted to add a 5th node, that node alone would contribute an additional like 15% overhead

Overhead on what?

@creamy-pencil-82913 following your HA K3s embedded etcd guide, each new server node added must cause more cpu,memory, and network usage, like more servers means more network traffic for control plane... and cluster would also have to make more decisions on where to schedule things, also etcd data sync between the servers, each nodes heartbeat health check, cluster services like coredns handling more load... i was wondering if you guys had a document showing a general like usage of these resources i mentioned at beginning and how these values change as you add server nodes for HA. if not i think i can just use prometheus to get these stats? regards -will