object: support batch scheduling for multiple different instances using multiple RequestSpec Object
Add batch scheduling for multiple different instances.
To support batch scheduling, the RequestSpec Object should supports multiple different flavor.
Problem description:
In liberty, RequestSpec Object added to encapsulate the request to launch one or more instances with the same flavor.
But in some scenarios, such as NFV, one VNF consists of multiple different instances usually, and all the instances must be launched successfully.
For now, nova scheduler only supports schedule different instances one by one.
And if one instance of a project failed to launch, all the launched instances of the project were uselessly.
For example:
one project includes 3 instances:
instance1: 8 vcpus
instance2: 6 vcpus
instance3: 6 vcpus
And the available resources are :
host1: 10 cpu
host2: 12 cpu
The scheduler may launch instance1 at host2, and then instance2 or instance3 will failed.
If the scheduler supports batch scheduling, we can schedule all the 3 instances at once : instance1 at host1, instance2 and instance3 at host2.
Blueprint information
- Status:
- Not started
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- wei
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- New
- Series goal:
- None
- Implementation:
-
Unknown
- Milestone target:
- None
- Started by
- Completed by