VMware Datastore Storage Backend
A VMware datastore storage backend*:
The VMware driver in Nova has to copy the image from the storage backend (filesystem for example) to the datastore to be able boot the VM.
We can get rid of this extra network transfer by having a datastore storage backend.
In this case, Nova will need to have an image handler** that is using the location of the image instead of downloading the bits to spawn an instance more efficiently.
Also, existing VMware customers will be able to create images using the "location(s)" (v2 api) to locate their existing files in the datastore.
The VMware datastore will be able to have its own implementation of the Store interface***, namely configure, get, add, delete, etc.
* https:/
** https:/
*** https:/
More information about datastore: http://
Blueprint information
- Status:
- Complete
- Approver:
- Mark Washenberger
- Priority:
- Medium
- Drafter:
- Arnaud Legendre
- Direction:
- Needs approval
- Assignee:
- Arnaud Legendre
- Definition:
- Approved
- Series goal:
- Accepted for icehouse
- Implementation:
-
Implemented
- Milestone target:
-
2014.1
- Started by
- Arnaud Legendre
- Completed by
- Arnaud Legendre
Related branches
Related bugs
Sprints
Whiteboard
Arnaud,
This looks good. However, I have few concerns regarding the life cycle of the instances created from the template. not so much with the glance changes:
.1. when booting an instance with multiple data volumes how will you map the volumes into Cinder?
.2. will you ignore the flavors
.3. how will you handle snapshots
-Yehia
Yehia -- I have a created another blueprint to track the OVA work. I move your comment to the other bp (https:/
-arnaud
Gerrit topic: https:/
Addressed by: https:/
VMware Datastore storage backend
Gerrit topic: https:/
Gerrit topic: https:/
Addressed by: https:/
Skip unconfigurable drivers for store initialization
