Enable secondary indexes
Summary
Secondary indexes are theoretically implemented, but before we enable them, we should:
- Thoroughly test their performance
- Improve the way we document/communicate how they work
- Consider only enabling secondary indexes on fields that are part of the key (which would increase consistency and performance)
- Consider not allowing secondary indexes on very large streams
- (Re)consider the
normalize
option on the@index
schema directive
Risks and challenges
TODO
Involved components
pkg/schema
pkg/codec
engine/driver/bigtable
Proposed changes
TODO