Details
-
Task
-
Resolution: Done
-
Major
-
OADP 1.2
Description
Potentially seeing issues w/ volsync running more than 10 volumes at a time in https://issues.redhat.com/browse/OADP-969
Definition of the problem...
Batching. Batching can help optimize backups and restore for performance as opposed to:
- as 1.1.1 running b/r in serial
- as proposed initially in 1.2 to ungovern the number of b/r at once and have openshift govern the # of processes, tasks, containers, pods.
The performance team in the context of Bell labs performance came up w/ an optimal 12 volumes as a batch size.
https://docs.google.com/document/d/1kPKo46_McEkj0Fu9hdkTxiVDnTh00forFQnscTgiEgg/edit
A potential solution could be implemented in the following.
- can be done in velero
- this may be slow and too slow for oadp-1.2
- can be done only in the data mover plugin
- we control the work here.
If we do this on the data mover plugin:
Attachments
Issue Links
- depends on
-
-
- New
-
-
-
- Closed
-
-
-
- Closed
-
- relates to
-
-
- New
-
- links to