Im trying to understand capacity planning for a putative Cassandra network. I have brought up a test group using a mixture of hosts (RHEL, Win64, etc) and it seemed fine. Im wondering how feasible this sort of thing is long-term. IE how do I define the smallest acceptable node? Whats the effect of adding large, powerful nodes to an installation with smaller nodes?
Im trying to understand capacity planning for a putative Cassandra network. I have brought
Share
Mixing different platforms (Windows/Unix) is not supported. It might work, but it probably will bite you at some point, and it’s a bad idea anyway — the more homogenous your environment, the less factors you need to consider when troubleshooting.
Cassandra does allow you to load balance across hosts of different capacities; see the operations wiki page.