In a recent blog entry I pointed out several wrong and/or suspect predictions made by industry pundits. But they're not always wrong. Some predictions seem so simple that anyone could have made it, at least sitting here in the future it can seem that way… but sometimes futuristic soothsaying can be correct, and quite useful to know.
While I was cleaning out my closet I came upon the 1996 conference proceedings of Gartner's Getting Your Data in Shape event. I thought it might be interesting to see what Gartner was saying back then, and what has transpired since. And guess what? Gartner did a very good job predicting the future. Now I don't want to be seen as a shill for Gartner - and sometimes I'm sure they are not as prescient as they were for this particular event - but having a dedicated team of analysts investigating a field and making informed, reasonable projections can be helpful to IT organizations.
In the next few paragraphs I will examine some of the forward-looking projections I found in this set of proceedings. Keep in mind that Gartner applies a probability to its predictions -- sort of a way of hedging their bets. Generally speaking, the Gartner analyst feels very comfortable about anything with a probability over 0.7... and a 0.9 is almost certainty.
For the next ten years, the extended relational model will remain the most viable platform for generic data management (0.8 probability).
This prediction was spot-on. The extended relational products (DB2, Oracle, SQL Server, and to a lesser extent Sybase) are the most commonly used DBMS products for most varieties of data management activities. You might scoff at this prediction and say something like "that was an easy call" - but to be confident of such a prediction over a 10 year span is impressive. And OO was threatening to be a big disruptive force in DBMS back then.
Although repository products offer help with the data warehouse metadata issues, by 2000, vendors will not provide a robust solution until distributed repository architectures with specialized "subrepositories" are delivered (0.8 probability)
Not bad here either. Remember, this planning assumption was made during the age of the big "R" Repository - Platinum technology was touting their Repository (which was a merge of the Brownstone and RelTech repositories), Microsoft was making noise about theirs, and repositories were being viewed as the glue that holds the IT infrastructure together. Of course, the big "R" repository has basically died and no real "robust" solution has ever been delivered. CMDB seems to have supplanted repository as the term du jour in this area.
Specialty data warehouse DBMS products will provide support for predictable queries appropriate for specific DSS applications only, and thus will fail as an enterprise data warehouse by 1997 (0.8 probability)
For the most part, this has come true, too. The specialty DBMS products (e.g. Red Brick and Arbor) are either gone or surviving only on the periphery.
Organizations should plan to spend at least 2.5 times their legacy systems management budgets on distributed systems management through 2001 (0.8 probability)
This statement was made at a time when the general consensus was that client/server implementations would be a lot less costly than mainframe implementations. And though Gartner doesn't use these terms specifically, they are basically saying that management of distributed systems is complex and costly. Another good call.
The maximum-sized database in production will always outpace the maximum manageable database size.
This one does not have a probability because it was drawn from a chart that graphed database size versus manageability. I've been quoting this nugget of information in my presentations for some time now. And it is true. There is always an implementation out there that is straining against the limits of what can be done today - and managed today. I think it means something like this: it would take longer to recover this database than to rebuild it from scratch so we won't back it up. But once again, a good piece of knowledge from Gartner.
Overall, the information in these proceedings would have been very worthwhile to an IT department back in 1996 as it worked to establish budgets and plans for the future. So, skepticism is a virtue, but sometimes analytical "guesses" can be helpful... just don't bet your career on them!