Feature #4613
closed
Allow bucket data to reside in a separate pool to object data
Added by Neil Levine about 11 years ago.
Updated almost 11 years ago.
Description
As an adminstrator, I want to allow the data about the contents of a bucket to be served from a SSD pool while leaving the object data in an HDD pool, so that I can provide a fast(er) response to users querying the contents of a bucket.
This would be especially useful for us, perhaps even a must. We have 200-250 million files split in a number of containers and we regularly do container listings, so I'm worried about a spindle performance for this. Note that Swift can you easily do that with Swift.
- Target version set to v0.67 - Dumpling
Implemented this one for dumpling. Made it possible to define different data and index pools for a bucket, and give users some way to control which pools to use for placement.
Specifically, the (new) region structure will have a field that will hold a list of placement targets. Each zone will have a list of placement targets (or rules) and how they map into data and index pools. There will be a default placement target, but users will be able to specify the placement rule they want for the bucket through the S3 bucket creation api.
It will be possible to limit use of placement targets to specific users.
- Status changed from 12 to Fix Under Review
- Status changed from Fix Under Review to Resolved
Also available in: Atom
PDF