Category: Database In-Memory

Last week I spent some time beta testing Oracle Database 12c RAC on docker (more on this in a future post) and decided to take a trip down memory lane and play around with Oracle Database In-Memory.

One of the things I looked at was the new FOR SERVICE sub-clause of the INMEMORY DISTRIBUTE clause.

ALTER TABLE customers INMEMORY PRIORITY HIGH DISTRIBUTE FOR SERVICE IM1;

This new sub-clause allows you to control exactly where an object (table or partition) is populated into memory in a RAC environment based on where a database service is allowed to run. If the service is stopped then the objects distributed for that service are automatically removed from the IM column store(s).

If you recall each RAC node has its own In-Memory column store (IM column store). By default, when a table is populated into memory in a RAC environment it will be automatically distributed across all of the IM column stores in the cluster. That is to say, a piece of the table will appear in each RAC node, effectively making the IM column store a shared-nothing architecture.

This is not always desirable, especially if you only run certain applications or workloads on a subset of RAC node.

The video below demonstrates how to use the new syntax and what gv$ performances views you should look at to monitor what is going on in the IM column store in a RAC environment. You’ll find a copy of the script I used in the video at the bottom of this post.

PLEASE NOTE: That the service names are case sensitive, so make sure you keep them consistent. If your service name is defined all in lower case and then you specify the name in upper case in the ALTER TABLE command, you will not get an error. Remember this command is only changing the INMEMORY attribute on the table. It’s only when we begin populating the table into memory will Oracle discover that the service you specified doesn’t exist, i.e. doesn’t match the case of the service defined.

Continue reading “Controlling where objects are populated into memory on RAC”

My favorite marquee feature in Oracle Database 12c is Database In-Memory. With the introduction of Database In-Memory in 12.1.0.2, data can now be populated into memory both in a row format (the buffer cache) and a new in-memory optimized column format, simultaneously.

The database maintains full transactional consistency between the row and columnar formats, just as it maintains consistency between tables and indexes. The Oracle Optimizer is fully aware of what data exists in the column format and automatically routes analytic queries to the column format and OLTP operations to the row format, ensuring both outstanding performance and complete data consistency for all workloads without any application changes.

So, what can you expect from Database In-Memory in 12.2?

Continue reading “What’s New With Database In-Memory in 12.2?”