There are some key features that make Oracle attractive to an enterprise that DB2 just doesn't do. Why didn't IBM try to add those features to DB2 and then the IBM shops might choose DB2 over Oracle?

For example, Oracle has an active-active shared everything environment in Real Application Clusters (RAC). That makes it very attractive. DB2 HADR is a shared nothing, essentially active-passive solution. However, a little product called Gridscale can make DB2 work in an active-active manner like RAC so why don't IBM buy it and resell it? They talk about it in their magazines!

Oracle partitioning is another key feature that can be mapped quite easily to existing DB2 functionality, and although IBM claim they can match it, is it really there when used at scale?

I still remember all of those middleware and federated database pitches.

When will we get some real isolation of massive applications from the programs that manage the writing of data to disk?

"Investing a modest chunk in EnterpriseDB is a way of having somebody else test drive a new strategy for carving a wedge inside the Oracle market"

I don't really buy this. EBS's ability to take market share from Oracle has already been found wanting. They're well known for have ploughed their way through lots of VC cash before. Rather than being a worthwhile punt, surely this latest deal is a lifeline to the troubled database supplier.

While you're trying to spin this, you still fail to mention IDS. aka Informix.

Now why is that?

Yes there is a "free" version from the IIUG. (International Informix User's Group) [www.iiug.org]. Sure there are limitations placed on it.

However, even with a throttled engine, you're still going to get a lot of product to play with, and you will still out perform the "open source" products.

To use your point, you're running an up and coming web site. You have limited budget and needs. IDS from the IIUG would meet your needs and allow you a decent amount of growth. Once you reach a certain size, you should be able to afford the whole kit and then still grow while on the same platform.

IDS *is* the industry's dirty little sekret. The sad truth is that there are many companies that will not talk about their use of IDS because it represents a competitive advantage.

But hey! The author is an "expert" with his 15 years in the industry. Why don't you talk with people who have 20+ years in this industry? I'll be hitting 25 this year, in Sept. I've worked with a lot of different databases, in a lot of different situations.

As the author says, its IBM's VC not their IM pillar that did this.

IBM is a huge organization and well, they don't always communicate. I'll wager that the bulk of the company isn't even aware that IBM bought Informix, 8 years after the fact.

I cover most of these companies extensively, e.g. on DBMS2. To summarize:

1. Postgres, especially in EnterpriseDB flavors, meets most needs for most organizations. That doesn't mean you should port your existing apps, although they do their best to make that easy in the case of Oracle or MySQL. But unless you're so small that it REALLY makes sense to consolidate vendors, or even to have a complete single-instance company-wide database, Postgres is apt to be a very good alternative for new projects.

2. The same is true, to varying degrees, of most other DBMS one could name.

3. The above article draws a false dichotomy. Going after the market that's willing to adopt products like MySQL is not contradictory to going after the markets that currently prefer Oracle.

4. Enterprise uptake has been slower for EnterpriseDB than Web/SaaS/ISV uptake. They even fired some enterprise salesmen late last year. But they do, at least under NDA, have a bunch of big-name enterprise customers.

CAM