Ask Your Question

Revision history [back]

Required Hardware for a scalable cloud?

Hello All,

I am looking at building a scalable Openstack cloud, and am wanting it to use as much commodity hardware as possible. I want to try and stay away from big end suppliers like NetApp etc, as these costs seem to be a hindrance more than provide many benefits. Therefore, please could someone let me know their thoughts on what I was thinking, and whether I should re-adjust/manage expectations?

Management Nodes- This cluster will be two hyper-V servers. These will host any virtual machines for Openstack and management servers. These will use local SSD disks, with HA being done by duplicating roles on each server.

Object Storage Cluster - This will be three pizza box servers with loads of SATA drives (high capacity). These will have 1Gb connectivity and will be used for all object storage.

Block Storage - The block storage system will be made up using Windows Server Scale out file server. I am thinking two nodes to begin with, with a mix of local HDDs. 10Gb connectivity. These will include some SSD and SAS drives.

Compute Clusters - These will be dual xeon, plenty of memory and SSD drives for OS/Config. 10Gb connectivity.

Does this look like an acceptable configuration? Or would you recommend I change the way things are setup?