Re: database market share 2003

From: Darin McBride <dmcbride_at_naboo.to.org.no.spam.for.me>
Date: Thu, 03 Jun 2004 14:37:53 GMT
Message-ID: <5JGvc.659814$Ig.278119@pd7tw2no>

Noons wrote:

> Darin McBride allegedly said,on my timestamp of 3/06/2004 10:18 AM:
>

>> I think Noons' point is that it's unfair to compare Oracle (Linux,
>> Unix, Windows) against DB2 (Linux, Unix, Windows, AS/400) since the
>> domains are different.

>
> Nope.

Then, please ... explain in plain English for us obvious idiots.

>> Larry wonders why it's fair, using the same logic, to compare Oracle
>> (Linux, Unix, Windows) against MS SQL Server (Windows) since the
>> domains are different.

>
> Nope.

Oh, I'm pretty sure that's Larry's question, based on the fact that the above is your implied (but not intended, it seems) point.

>> all-HP, all-Sun, all-IBM, etc.), I would think that most database-using
>> shops are simply trying to store and retrieve data.  As long as that
>> works, the platform in use is only of as much interest as their budget
>> allows (i.e., a budget of $20,000 isn't going to get an AS/400!).

>
> So, if you are using a AS400 to run RPG applications inherited
> from your 20 year old System 36, WTF is IBM counting THAT
> as a DB2 license?

Is this a significant portion of the AS/400 market?

>> If they go out and purchase some hardware running DB2, or Oracle, or
>> SQL Server, that's what they bought.  So it counts for that product.

>
> Sure. the problem is that IBM is counting ALL (let me see if you can
> grasp the difference: I said ALL, it means THE TOTALITY OF) AS400 sites
> as DB2 sites. Which they are NOT, NEVER were and NEVER will be.

Again - is that significantly different than the reality of how many AS/400 sites are actually using the DB2 that comes with AS/400?

>> counted as IE users, for example.  However, what Noons has not proven,
>> or even attempted to demonstrate, is that this is a significant
>> distortion of the reality.  Fact is that they did buy DB2, although
>> reality may be that they didn't want to.  But is that a significant
>> portion of the AS/400 market as to render it misleading?

>
> Yes. ALL AS400 sites IS a "significant portion of the AS400 market",
> in case you have not noticed?

You're answering my question backwards. I'm asking about the portion of the AS/400 market which IBM is incorrectly claiming uses DB2. Is THAT portion significant? I'm not asking if IBM's claim is significant - you may think I'm stupid, but I'm not that stupid. That it is the entire AS/400 market is obvious.

We're talking about statistical significance here. We have three numbers we're dealing with:

The totality of the AS/400 market.
The subset of #1 which is actually using DB2, or, in reverse, the subset of #1 which is actually not using DB2. Obviously, one of these implies the other.
The overall DB2 market.

The questions are:

What is the ratio of 2:1? Is it statistically significant? If it is 50%, then that may be significant. If it's 95%, then we're just blowing a lot of smoke over nothing.
What is the ration of 2:3? Is this statistically significant? If we're changing IBM's numbers by removing AS/400 users that don't actually use DB2 from 37.6% to 37%, why are we having this argument?
Regardless of what the ratios are, it is obvious to the rest of us that you cannot be honest if you are asking to remove the entire AS/400 market. Honesty would require removing only the portion of the AS/400 market which is not actually using DB2, whatever that may be. You must concede, however, that doing that is not easy. It may be the most honest method of reporting, but it is not easy to do. A customer may buy an AS/400 not intending to use DB2, and then do so anyway (they are licensed to do it afterall). Or a customer may purchase Oracle for HP and then the project is cancelled - discounting this from Oracle's numbers is not going to be any easier. It would mean going to each vendor's customers, and verifying that each one is using what they paid for. Definitely honest, but is it going to produce significantly different numbers that would justify the expense?

I doubt it. Received on Thu Jun 03 2004 - 09:37:53 CDT