Ivan Bessonov created IGNITE-17338:
--------------------------------------
Summary: Implement RocksDB based hash index storage
Key: IGNITE-17338
URL: https://issues.apache.org/jira/browse/IGNITE-17338
Project: Ignite
Issue Type: Improvement
Reporter: Ivan Bessonov
Please see IGNITE-17318 for partial description of what needs to be achieved.
I expect that hash index records will have the following structure:
{code:java}
[ indexId | partitionId | hash | tuple | rowId ] -> []{code}
Fixed-length prefix should cover indexId, partitionId and hash value.
Searching rows effectively becomes a scan, but this is fine.
Hashing must be performed internally, hash function already presents somewhere
in the code.
Is far as I understand, PK is going to be implemented as a secondary hash index.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
