can we safely upgrade a cassandra cluster with more memory and cpu? -
all of our cassandra nodes vms. if load data , begin using system, can add more cpus , memory vm @ later date? cassandra automatically handle that? there changes need make?
this version 2.0.5
assuming virtualization layer allows add cpus , memory, there should no problem.
you may need adjust java heap settings (see conf/cassandra-env.sh
) allocate more memory cassandra, require rolling restart of each affected node (depending on virtualization layer, may need restart anyway add cpu/ram, , java requires restart detect additional cpus).
note best performance (and fault-tolerance) out of cassandra running on bare metal, not vms.
you may want consider upgrading cluster adding more nodes, rather increasing resources of each existing node - depends on use case.
see also:
- tuning java resources (datastax)
Comments
Post a Comment