Google Search » Database Underground | Sean McCown » TAG: DB2

August 29, 2006 | Comments: (0)

The Viper Mistake

TAGS: DB2

OK, as I'm sure every one of you read my review of DB2 Viper and hung on my every word, so it's only fitting that I clear up a mistake I made in my numbers.

The problem is I got it right in the preview, but somehow flubbed the numbers for the formal review.

It should have read that Viper's XML engine has been measured up to 7x faster than Microsot and Oracle, not 7%. So that was my bad, but it still doesn't change my original assessment. I still think that it would take a pretty decent XML load on the server for you to really feel the difference. Think about a single insert statement. You type it in and hit run and it returns instantly with 1 row... right? Now imagine that happening 7x faster. That doesn't really even mean anything to us. So there are 2 scenarios where I can see this really coming into play... the first is where we just described... in single inserts done by the thousands. In this case, you'll probably feel the difference in speed pretty significantly. The other scenario is in very large inserts. If you're inserting an XML file that's really big, and it's the bulk of the load, then you'll probably feel this speed increase as well. However, if the XML file is just a small part of an otherwise large insert, then it probably still won't make that much difference because it's not the bottleneck to begin with. So say you were inserting a 2GB blob with it then you probably won't be impressed with the speed as much as if you were inserting it by itself.

So I still say that the large portion of the market won't see mass returns from this. It does bring up another really good point though. I'm still really wondering what they've got in store for the future. The new engine will preserve the digital signatures of XML stored in the DB. And right now I don't think there are enough people passing around digitally signed XML files to really get a lot of use out of this new engine. What it does make me wonder though is whether they'll be able to do this with other types of docs in the future. Will we be seeing native Word formats, or Excel, or PDF? These are file types that people sign all the time and the native storage of the XML engine would be an excellent use in these areas.
Just imagine... a native Excel data type in your tables. Or like I said... a native PDF format.

Posted by Sean McCown on August 29, 2006 07:44 AM

May 01, 2006 | Comments: (0)

What is the Future of DB2?

TAGS: DB2

I often wonder who will ultimatly take the lion's share of the database market. Oracle and Microsoft seem to have a strong lead while IBM and Sybase seem to be lagging behind. And for some reason Sybase seems to be almost not even on the same course as the others.

So I was on the phone with one of the IBM execs last week, and I asked him why I never see DB2 in full production in companies I visit, and when I do, it's usually only in the process of being transitioned to Oracle or SQL Server. He was quite frank with me and said that IBM hadn't been as good at marketing DB2 as they'd like. I didn't really get whether there were any active plans to increase the marketing campaign or not, but he was very discouraged to hear that DB2 isn't more ubiquitous than it is.

The truth is that DB2 has some ground-breaking features that could take the database world to the next level. I think we can all agree that IBM has always pioneered database research. Their last couple versions of DB2 have brought some incredible advancements to the world of databases. The methods they use for updating statistics, caching execution plans, storing XML data, and providing row-level data compression are virually unmatched by any of the other vendors. So again, why are Oracle and Microsoft killing DB2 in the marketplace?

Sure, marketing... that's certainly a factor, but I think the other part of the answer is in the very nature of cross-platform software. Oracle suffers from this to a degree as well. See, DB2 isn't very windows friendly. I haven't found the code editors to be loaded with user-friendly editing features, nor have I found them to be rich in visual aspects either like having nice interfaces, etc. A perfect example can be found in the simple act of running a query. I haven't found anything in the editor that shows you how long the query took. In SQL Server, there's a space at the bottom of the query window that shows you how long the query took. In DB2, the execution time is shown in a separate pop-up window and it goes away as soon as the query is finished, so if you're not watching it, you have no idea how long it took. I ran into this problem recently when trying to compare the execution times of two separate import methods.

Another good example is in the pure lack of astetics in the GUI. I remember a very recent discussion with the editors of the magazine where we were trying to find a good screenshot to put with the preview of Viper, and what it came down to was finding the lesser of the evils. We never did find an eye-pleasing screenshot that was worthy of being printed. You don't have that problem with SQL Server, especially 2005. I think if IBM put some mroe work into their GUI, and gave us a good native development environment, they may be able to market it a little better.

That's really what it boils down to isn't it... astetics. I know if I'm going to spend 12hrs/day in front of a screen I want it to be something nice to look at... even if it's code. I also want the editor to help me as much as possible, and IBM really misses the mark on that one.
What I'm trying to say is simply this: I love the innovations DB2 brings to the table, but I want to stab myself in the eyes every time I have to do something with it because it's just a terrible interface to work with. So being cross-platform is great, but to make DB2 so generic as to be able to fit in all the platforms the same way just keeps it from being truly great on any platform.

I really don't know what the future holds for DB2, whether IBM will be able to market it better or not, but I'm fairly convinced that if they don't do something soon, MS and Oracle will get so far ahead of them, they'll never be a major player in enterprise databases.

Read my book reviews at:
http://www.ITBookworm.com

Posted by Sean McCown on May 1, 2006 05:11 PM