At CloudWorld 2022 it became apparent to the Database In-Memory team that many customers are unaware that Database In-Memory columnar formats are supported in Exadata Smart Flash Cache. This is in addition to the In-Memory (IM) column store that can be allocated in the database instance’s System Global Area (SGA) on the compute nodes.
In fact, starting with Oracle Database version 19.8, it is possible to set the INMEMORY_FORCE parameter to CELLMEMORY_LEVEL and only make use of Database In-Memory columnar formats in the Exadata Smart Flash Cache without having to allocate any memory in the compute nodes. This can be a big advantage for customers that cannot afford to allocate compute node memory to the IM column store but still want to take advantage of the big analytic performance improvement that Database In-Memory enables. Since the Exadata storage servers automatically manage what gets brought into the Exadata Smart Flash Cache in Database In-Memory columnar format, or cell memory, there is no need to figure out which objects to populate. It is all done automatically.
If this is news to you then there is plenty of information available about how this works and what needs to be done to take advantage of this feature. Check out the links below if you are curious and want to find out more.
Documentation:
Previous blog posts:
- Database In-Memory and the Autonomous Data Warehouse
- Database In-Memory Columnar Format in Exadata Flash
- How to Determine if Columnar Format on Exadata Flash Cache is Being Us …
- Using Database In-Memory to Supercharge Exadata
- Columnar Formats in Exadata Flash Cache
- New Database In-Memory Features in Oracle Database Release 12.2
Roger MacNicol’s posts that are now available on the Database In-Memory blog:
- SmartScan: What’s New in 12.2 CELLMEMORY, Part 1
- SmartScan: What’s New in 12.2 CELLMEMORY, Part 2
- SmartScan: What’s New in 12.2 CELLMEMORY, Part 3
Original publish date: 11/1/2022