Is it possible to have separate data file(s) per bucket?
I would like to have the data files for each bucket on different volumes so that they can be tuned for the workload expected for that bucket. Is this supported and/or do you think this is a valid tuning approach?
In Couchbase 2.x you can only chose the location (volumes) of:
- the index
- the data files
and based on your question, you already know that in production it is a good practice to use different volumes.
Also you may have seen that the data files are already in multiple files, but on the same location on each node of the cluster.
Here some interesting information about files and compaction:
- Compaction magic in Couchbase Server 2.0