RLV Data Store for Write-Optimized Storage

Use the new row-level versioning (RLV) data store in your simplex database to perform row-level updates, inserts, and deletes, in real-time. When a table is registered for storage in the RLV data store, multiple users can write to different rows of the same table concurrently.

For information on architecture, configuration, blocking, locking, monitoring, merging, and troubleshooting row-level versioning, see the new manual Administration: In-Memory Row-Level Versioning.

SAP Sybase IQ 16.0 introduces a new data store:

• RLV Store – The in-memory data store optimized for high-performance row-level updates. The RLV store acts as a staging area for write events. If a table is enabled for row-level versioning, then all LOAD TABLE, INSERT, UPDATE, and DELETE events write directly to the RLV store. In-memory data in the RLV store is periodically merged into the IQ main store. You can set parameters for automatic merges, and you can merge on-demand.

The IQ main store is optimized for high performance queries, rather than table writes. The RLV store adds high-performance write capability to your database. You can use row-level versioning in conjunction with your existing insert mechanisms such as bulk loading. For example, your applications can make real-time updates during business hours, while overnight a bulk load operation bypasses the RLV store and loads updated data directly to the IQ main store.

If a table is RLV-enabled, SAP Sybase IQ table DDL behavior is unchanged. DDL events on a table (adding a column, for example) require an exclusive lock on that table.

See these topics in the SAP Sybase IQ 16.0 documentation:

• Administration: In-Memory Row-Level Versioning > About In-Memory Row-Level Versioning
• Administration: In-Memory Row-Level Versioning > Appendix: SQL Reference