If I am running a kopia client against a remote kopia server, what type of loads are put on either side. If the kopi server is a older or slower model, than what the client has, will the client still do the heavy lifting like splitting and hashing?
The servers CPU specs will probably matter when it comes to repo maintenance, since kopia server does this on his own. I am thinking of running a remote kopia server on a Intel D-1518 @ 2.20GHz which gets easily outperformed by the i7 on my MBP 16".
FWIW, I run my Repo Server on an Intel Core2Duo 2.4 GHz from 2006 (yes, that old).
From what I can tell looking at the graphs while a snapshot task is running, RAM (only 4 GB here), disk speed and network bandwidth seems to be the real choking points. CPU is toggling between 50% and 75% use.
I can’t tell you anything about repo maintenance though.
Yeah - the kopia client performs all of the heavy lifting. The server doesn’t get too busy, not even when performing full maintenance work. On the first snapshot network-throughput will be the main choke point. Subsequent snapshots will suffer from high latency, like on wireless networks vs. LAN ones.
My kopia server runs on a virtual guest with only 2 vCPUs and 4 GB RAM and it’s working absolutely smoothly. I added a fast cache for metadata caching, though. This helps with drive related latencies, when performing subsequent snapshots.