1570462507
explode metadata key/values to their own tables/rows (based on type). build a key string using dot notation similar to other nosql db and filter based on that. Blueprint: sqlalchemy-metadata-query Related-Bug: #1093625 Change-Id: I2076e67b79448f98124a57b62b5bfed7aa8ae2ad
2.5 KiB
2.5 KiB
Choosing a database backend
Selecting a database backend for Ceilometer should not be done lightly for numerous reasons:
- Not all backend drivers are equally implemented and tested. To help you make your choice, the table below will give you some idea of the status of each of the drivers available in trunk. Note that we do welcome patches to improve completeness and quality of drivers.
- It may not be a good idea to use the same host as another database as Ceilometer can generate a LOT OF WRITES. For this reason it is generally recommended, if the deployment is targeting going into production, to use a dedicated host, or at least a VM which will be migratable to another physical host if needed. The following spreadsheet can help you get an idea of the volumes that ceilometer can generate: Google spreadsheet
- If you are relying on this backend to bill customers, you will note that your capacity to generate revenue is very much linked to its reliability, which seems to be a factor dear to many managers.
The following is a table indicating the status of each database drivers:
| Driver | API querying | API statistics | Alarms |
|---|---|---|---|
| MongoDB | Yes | Yes | Yes |
| MySQL | Yes | Yes | Yes |
| PostgreSQL | Yes | Yes | Yes |
| HBase | Yes | Yes, except groupby | No |
| DB2 | Yes | Yes | No |