Implement load-balancing and failover in SQL-based backends using HAAlchemy
Ceilometer has pluggable storage backends including SQL-based that use the SQLAlchemy library for connection and session management. The conventional way is to use HAProxy between the controller node and the database cluster to load-balance database requests, which requires additional configuration.
Instead, the proposal is to use HAAlchemy (http://
Blueprint information
- Status:
- Complete
- Approver:
- gordon chung
- Priority:
- Undefined
- Drafter:
- Rohit Jaiswal
- Direction:
- Needs approval
- Assignee:
- Rohit Jaiswal
- Definition:
- Superseded
- Series goal:
- None
- Implementation:
- Unknown
- Milestone target:
- None
- Started by
- Completed by
- gordon chung
Related branches
Related bugs
Sprints
Whiteboard
disclaimer: i've never used haalchemy... that said this seems like a sane solution... my question is this an optional path where it still allows users to use HAProxy? also, what's the intended scope of this? events only? -- gordc (28.05.15)
gordc: We can make it optional so that HAProxy can still be used, however HAalchemy seems like a better solution..This should cover both samples and events.
i see. you can open a spec i guess. not sure how this is any different from sharding in sqlalchemy -- gordc (10.7.15)
not applicable, deprecated for http://