I'm happy to announce the next Sophia release.
Most notable changes are the new storage model and storage architecture.
New storage model works with rows. Each row holds fields of arbitrary type, like string and numbers. This allows significantly reduce
storage overhead. Eg. numbers are stored without overhead inside a row. New scheme also able to work with secondary indexes.
The new storage architecture, which now has strict O(1) complexity for read, write and range scans.
I hope this will be the main distinctive Sophia feature (as in origin idea).
Changelog:
- Next generation storage architecture, O(1) by default
- New storage scheme: row model, typed fields
- Storage optimizations: zero-cost numeric store, pluggable meta-data
- Support for secondary indexes
- O_DIRECT support
- Separate compaction settings per database/index
- Performance and profiling
- Automatic key expire
- Custom comparator, limits
- Lots of refactoring and tests in production
Documentation is available on the http://sophia.systems website.
Thank you