I don’t know if you caught Larry Ellison’s announcement on June 10, 2014, but Oracle has been hyping their new Database In-Memory tech and I figured I’d summarize what I’ve found about it so far.
The Oracle press releases speak to how revolutionary Database In-Memory will be and how it will benefit customers. So what is it exactly is it? One of the main uses for Oracle Databases is obviously to store data. But one of the things enterprise organizations also look to do is to then use that database to perform analyses.
Storing data can be highly efficient in Relational Databases because of optimizations and the ease of storing data in rows. Analyzing data in a Relational Database can be time/energy intensive because you’re not just interested in the relationships between data in different rows, but also the same column. Generally that requires copying the dataset out of the database and into another system, or storing it in a different manner increasing either I/O usage or storage usage.
What Oracle appears to be doing is modifying their database so that it can be accessed either via the typical RDBMS row paradigm or via a columnar paradigm. The initial announcement implies that all of this is performed in memory so that: a) no additional storage is used; b) they use their optimization algorithms on the most used data to store in-memory in order to increase throughput/response time without storing all the data itself in-memory. Another of the major claims is that the end users don’t need to do anything on the application side to implement this increase in speed/efficiency, so it should just work better for end users.
I don’t think it’s really going to affect those dealing strictly with Identity and Access Management unless there is a client who does extensive analysis with BI Pub or maybe EBS suite. On the other hand, Oracle asserts that with no changes on the app side – and once the DB makes the change – it’s seamless to the clients. In theory that means we shouldn’t be affected with any of the FMW apps, but as with any new functionality I suggest checking expectations until we actually encounter it in use.
The best reference I’ve found so far on the technical details can be found in this PDF from Oracle.
Thought I’d give you a quick primer on what it’s doing since none of the press releases seemed to go into what it actually does behind the scenes.