Gartner: Are Databases Dead? DBAs to Follow?
First of all, let me say that it is the job of an analyst to be provocative. With this position, these guys do a good job on that front. But what of the meat of their notion: will databases disappear? Well, no, they won't. If you read the article closley even these guys don't actually say that databases and DBMSs will go away.
Secondly, what of the notion that there is some data that need not be persistent in order to be accessed and processed? This is actually interesting and a probably future development. And it might be further along the road to being a reality than you might imagine. Michael Stonebraker, the database pioneer behind both Ingres and Illustra, is involved with a project focused on data management for streaming data, using a SQL variant called StreamSQL. He has founded another company, StreamBase Systems to commercialize the technology. Given Stonebraker's track record, I wouldn't bet against him.
Some of the other ideas that are highlighted in the Gartner article are somewhat amusing (at least to me). First off, they talk about RFID as a technology that can enable database-less inventory. No time soon guys. RFID tags do not communicate over long distances. So what are ya gonna have to do, walk up and down every nook and cranny of your warehouse whenever you want to run inventory processing? Doubtful.
Read the comments posted back at the site housing the Gartner article, too. There is good one there asking about how you reconcile your inventory if your warehouse burns down... if the warehouse is the database it burned down, too, right? And there ain't no disaster recovery backup of your entire warehouse stored anywhere.
Another interesting tidbit is that these guys once again trot out the old DBAs are toast chestnut. Here, from that article, is the section I am referencing:
Feinberg and Friedman advised the DBAs in attendance that they should be thinking about which of two new roles they'd prefer; that of a repository administrator or that of a data service administrator.
So these guys think the captial "R" Repository is coming back because databases are dying and DBAs should think about managing the Repository? I wouldn't bet my career on that. Managing and working with metadata is a viable choice, but repository administrator? And what about data service administrator? Sounds like the old drumbeat of "process" over "data" being pounded yet again. I don't buy it.
So take the time and re-read the article again. Bypass the hyperbole of extinct databases and non-existent DBAs. Processing the clickstream and processing non-persistent data are interesting and useful ideas. Can't we just talk about that instead of roiling up the masses with exaggerated calls of obsolesence?
