[CASSANDRA-13214] Data versioning for rows

• Type: Improvement
• Status: Open
• Priority: Normal
• Resolution: Unresolved
• Fix Version/s: 5.x
• Component/s: Feature/Materialized Views, Legacy/Local Write-Read Paths
• Assignee: Unassigned
• Reporter: Bhuvan Rawal
• Created: 12/Feb/17 17:11
• Updated: 07/Mar/23 10:53

Description

Row level data versioning can be implemented in cassandra in two possible ways :
1. Specify the last clustering column as a timestamp and all the previous data is merged into latest one. User can specify versioning while creating the table.
2. Create a materialized view with last column as timestamp without really touching the base table. Although a bit different from traditional MV as here it will contain data as versioned and in MV read before write complete partition data can be sent.

This will help people maintain historical records versioned without current client level challenges - consistency challenges etc, and rather doing at server level using things already implemented in place at cassandra level - materialized views, partition locks for MV update.