Persistent disk strategy backed by local disks disk offering

[poblin-orange]

Local disk offering for persistent disk is an optional strategy which may be useful in some scenarios, to provide interesting latency/bandwidth and cost saving in some uses cases (typically Mesos / RDBMS / ... uses cases).

The current implementation of the persistent disk using local disk offering causes random issues when attaching disk since local disk may not available to vms that need to use them.

Should design a mechanism to prevent this issue, such as

• try to colocate vm creation on same host as the local persistent disk. See related bosh director issue https://github.com/cloudfoundry/bosh/issues/945.
• snapshot / recreate persistent disk on the fly to match an existing vm location
• use xen supported live migration a "fake vm" to which the persistent volume is attached.

[tahiri-ahmed]

In the recent versions of Cloudstack (now version 4.9), this feature is supported by Cloudstack. See http://docs.cloudstack.apache.org/projects/cloudstack-administration/en/4.9/storage.html#creating-a-new-volume

[poblin-orange]

Hello, my understanding is that this feature is available for cloudstack operator, and could be used to configure a disk offering for ephemeral disks (not for persistent disk, as data availaility in case of host failure is not garanteed) This issue tries to deal with persistent disks. Not sure however that we can solve this @ cpi level. Maybe a solution like Bosh rackHD CPI, or scaleIO would be a better fit.