In-memory Databases Do Not Excuse Fat, Bloated Applications
In-memory databases are great, but please no bloat!
InfoWorld reports that Hasso Plattner, SAP's CEO, is going to push for increased adoption and use of in-memory databases. He states that using in-memory databases will create a revolution in new applications, especially in the area of business analytics.
Oracle's Larry Ellison disagrees, of course, calling it a "complete fantasy" that in-memory databases will replace SQL relational databases.
I agree with Plattner on this one. New developments in in-memory databases and in large, distributed database technologies like Hadoop are solving both speed issues as well as massive data issues. I really don't think Plattner is planning on replacing SQL with in-memory databases, just augmenting them where it makes sense.
However, this is a technology issue. SQL databases, in-memory databases, distributed databases, or even flat file databases lie so far under the application that discussions of this bore the business user to tears. You can boil this all down to: databases are getting faster, and bigger, and your IT department may ask for a few bucks to get you there. The CEO of SAP is going to talk about this during his opening keynote at SAPPHIRE, and I would guarantee that the excitement in the room will not be palpable.
However, regardless of how techie this is, it does affect us. Here's the real issue about this, from my point of view: with new developments in faster technology comes application bloat, as the developers no longer have to work around speed limitations. The challenge is to increase usable functionality (as Plattner is suggesting) without also introducing fat, bloated code. We are all familiar with the thrill of getting a brand new and much faster system, only to have our next software update chew it all up. One of the key benefits of an old work horse like Oracle JD Edwards World is that it was written when computers were a lot slower than they are now, and therefore it was written very efficiently. It would be very bad if companies were forced to implement in-memory databases just to keep their performance levels stable when they upgrade. What we can do to avoid this is to challenge our respective vendors when they send their sales reps out selling these new applications, and ensure that we add performance metrics to our evaluation process.
So, go for it, SAP! Give us lightning fast analytics! But please don't bloat the software any more than it already is!