Description
What is the problem?
Our documentation for the merge command currently says:
Merging data will deduplicate similar keys in backup files being merged together in order to create a single smaller backup file
Although this was true for the ForestDB and SQLite backup formats, it is not true for Rift. This is because the Rift data file is append-only, so when we "replay" the documents from each backup they are just inserted at the end. It wasn't true for SQLite because the document was stored in the index and each document had only a single row in the index.
What is the solution?
We should make this clear. Note that some space will be saved as there is an overhead for each backup, but this will likely be negligible.
Attachments
Issue Links
- relates to
-
MB-58030 [CBM] merges should deduplicate documents in the Rift format
- Open