The Cloud VPS project currently uses local storage on the labvirt hosts for virtual machine runtime images and an NFS cluster for off-VM persistent storage. This model requires that each labvirt (execution node) in the OpenStack cluster have locally attached disk sized to match the RAM and CPU of the host. It also means that the uptime of customer VMs are directly tied to the uptime of the underlying exec node. This complicates system maintenance, limits our ability to fairly distribute resources, and introduces a single point of failure for each and every VM.
- roughtly estimated 6-9 month project
- initial: 1.5 root, 100K hardware investment
- sustaining: .5 root, 20K hardware investment