Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

Another one asking about hardware

We're in the beginning stages of using some older hardware to start a migration to node based storage stack. Would there be large ramifications in using non matching hardware for the nodes?

Thats hard to say. The rule of thumb is the slowest osd will limit the performance in a pool. This is determinated e.g. by cpu, drive and network speed.

Also it is better to have unified servers and media for having spare parts available.

Well to test the waters I gathered up 3 older 12 bay R515's -  R700 raid controller, 12 core with 64G ram each. Should be a good starting point to get comfortable with the system before running it in production. Our intent is to expose an iSCSI target to our main vmware stack.

A question I have is if anyone has run into issues using large drives (>8T) with these H700 controllers. Anyone know the maximum size drives that will play nicely? Or does it matter since I wont be using the controller to build an array in the first place?

Also when sizing these drives, is it generally better to run with a higher number of smaller drives? Say 12 6T-8T drives vs 6 16T-18T drives? Any benefit to either way?

End state I will be running spinning drives with SSD for cache and journal in an attempt to keep cost down during our initial trial. These first nodes will be used primarily for our centralized backup repository then eventually be added to for production purposes.