am i correct in that coordinator nodes dont really require much resources? they arent even in the query path right?
is choosing t4.micro instances (1gb ram, 1 vcpu) good enough for coordinator nodes?
am i correct in that coordinator nodes dont really require much resources? they arent even in the query path right?
That’s correct, the coordinators are used to store some metadata of the cluster and they are not in the write/read path.
is choosing
t4.microinstances (1gb ram, 1 vcpu) good enough for coordinator nodes?
You might want to add some more RAM to the machine and do some testing, I’m not sure if 1 vcpu is enough (given that some OS processes will also be running on that host). You probably want to test how long it take to replace a coordinator and how long a recovery of the cluster takes.
1 Like