I begin to understand some things that happens in the past two years.
Why IBM buy Informix? They have DB2 in the relational market. Where was
the benefit? And in those years, what would they do with U2? I thought
that it was on the packet and they will leave it in a few years.

Today i have discover the BETA new DB2 product from IBM. It's name, DB2
VIPER. Surprise, a NATIVE XML database. Something similar to
TigerLogics from RainigData. I am quite sure they have used the U2
technology, as Raining Data uses the Pick technology.

May we continue using "our old technology" (basic and talking about
GUIs, transfers, etc.) or must we use "our old technology", with this
new wave. I am sure we are much more prepared that the "relational"
people in order to use this model. May be they name it Native XML, but
at the end, it is OUR database model.

What do you think about this?

I wait for your opinion.

IIBM Canada has been working on that for a least several years. That
group probably doesn't know that there are two IBM product called
UniVerse and UniData.

Even within PowerPC I have found that the group that handles the Power5
processor doesn't really know what is going on with the 970 which Apple
calls a G5.

The basic fact is that IBM bought the U2 line because it is very
profitable.

Henry


jra wrote:

If you can't kill an opposition, you should head it up. And then kill it. I
believe that's what IBM is doing to U2. On the other hand at the same time
they extract best features out of it and integrate into their mainstream
product.

I wouldn't wonder if one day I see multivalue features presented in MS SQL,
Access, etc.

Nick

"jra" <frelance (AT) sarenet (DOT) es> ???????/???????? ? ???????? ?????????:
news:1135291005.133918.170810 (AT) g43g2000cwa (DOT) googlegroups.com...

"Nikolai Lukin" wrote:
Keep your eyes open for the term "MDX".

http://msdn2.microsoft.com/en-us/library/ms144884.aspx

T

Has Tony G. been the behind the scenes consultant to microsoft?

and . . . isn't this confirmation enough that the md term will catch
on in mainstream IT while the mv term will always be met with a Huh?

How often does one hear bison referred to as bison? Right, seldom.
99.999% of the time people refer to them as buffalo and so it goes with
our own bison. The sooner we start using the md term the sooner will
people know what we are talking about.

However, this who microsoft concept is still based on sql theory and,
for those of you who dd't know that, Teddy Codd is DEAD!

Henry Keultjes
Database Scientifics
Mansfield Ohio USA






Tony Gravagno wrote:

Tony Gravagno wrote:
I could be wrong, but I think this is an initiative to help make the
SQL Server OLAP facilities more integrated into the mainstream. It is
about multidimensional and not multivalue. The logical data models are
different for MD and MV. With OLAP (Multi-Dimensional) products, you
model a fact table and a bunch of dimensions. This would correlate to
how we might model a file with a multi-part key. Each part of the key
represents a "dimension" and we can then reference the data from these
dimensions in virtual fields.

OLAP cubes and related tools typically store aggregate data, where
files with multi-part keys in Pick typically have aggregate data as
derived data (virtual fields). Otherwise, these are used for similar
purposes.

The Pick/MultiValue model is wider than the Multidimensional models in
that the entire enterprise can be modeled handily.

Just in case I misunderstood the purpose of MDX, if I took us on a
tangent, my apologies.

Related to the Original Post, if the Viper team used any U2 code, IBM
hid that (at least from me) pretty well since I poked a bit (but never
asked that particular question directly). The XQuery project was
happening on the DB2 side of the house, seemingly, without any apparent
scheduled project of XQuery against U2. The U2 team seemed to be
working on SQL projects instead. This offended my data modeling
sensitivies just a tad. It seemed to me it would be smarter to do
XQuery against the XML-like data model and SQL against the Relational
model. I would have prefered that my exchange with IBM on that one
would have gone more smoothly as I did not show my most patient side.
I have a few questions for U2 marketing that I'll ask in the new year
and I'll add the question of the relationship between Viper and U2 to
the mix.

Cheers! --dawn

Responding to Dawn and Henry:

Nick was wondering about the possibility of "multivalue features" in
SQL Server. I pointed out that Microsoft is indeed building in
multidimensionality which we readily recognize as a multivalue
feature. My one line response didn't intend to imply that Microsoft
is implementing a Pick model. For many years in trying to explain the
Pick model we've used Excel as an example: "See the columns and rows?
What if you want more than one phone number in a single cell?" Enter
the third dimension of values and a fourth dimension of subvalues.
It's no surprise that relational people have also found value in this
concept. They also know that you can store data using relational
methods, but to _use_ data ala OLAP, you need to represent the data
more as a logical cube. I'm sure in their great wisdom the relational
world will eventually come up with something resembling the MITS
hypercube too - I know this already exists in data warehouses, they
just don't call it that... I don't know what they call it, maybe just
a cube.

This MV/MD thing is akin to a "is a" relationship vs "has a". SQL
Server _has_ multidimensionality via MDX, but SQL Server _is not_
multivalue.

No Henry, while I'm sometimes on the fence about using popular
terminology to describe ourselves I think in this case it will lead to
trouble. We are not relational, and calling ourselves such as I saw
recently on a MV webpage is misleading and asking for trouble.
Likewise we _are_ not multidimensional in the sense that mainstream
people are starting to recognize it, so we need not change our MV
moniker, but we do _have_ the same multidimensional features as MDX,
and I think it would be a good idea for the astute VAR to use this in
the sales cycle. (I would say use it in Marketing but we all know
that Pick people don't do Marketing. Anyway...)

With MV and MD: We _are_ MV and they _are_ relational. We both _have_
multidimensional. If we change the identity of the technology we use
to MD then we _are_ nothing, especially if the common notion of
multidimensionality changes over time in a way where we no longer can
claim to even _have_ that feature, let alone _being_ an example
implementation of that feature. We could fall back on being the Pick
model, but that has it's own issues.

I think for now it's fine to say something like "our application uses
_xyzDBMS_ which uses a multidimensional data model, used for decades
before the concept became popular in relational circles, but it is one
of many DBMS products in the family of multivalue database engines".
That gets out all the buzzwords, is 100% accurate. If you are talking
with someone who is so critical about the terms you use (where the
difference between MV and MD will make or break the sale), then you
lost the sale before you walked in the door and your vocabulary won't
change that.

TG@ no_i_dont_honkin_consult_to_microsoft_removethisNe bula-RnD .com

Dawn,

You are correct. MDX is a standard Multi Dimensional eXpression
language, and whilst originally introduced by Microsoft, MDX queries
can be understood & resolved by most mainsream OLAP engines

IBM purchased informix because it was up for sale. Informix a few
years before had purchased Ardent (the owners of Unidata and Universe)
because of a datawarehousing product called Datastage. The revenue
from this product had apparently greatly exceeded the sales from the
databases (datastage used a modified universe engine). The plan was
to port datastage to Informix Online. In the end the top exec's nixed
that idea and created Ascential and tood Datastage with them.
Everything else (Informix, Universe, Unidata, redback, etc) was sold to
IBM. IBM promptly discontinued the informix tools and continues to
this day to support the U2 products.

Just recently it appears that IBM has purchased Datastage and has
merged it into their websphere product.

I was just pining about this in the Openqm forum as I am now being
forced to justify the pick model. I needed to show companies that were
successful with the pick model/application and Datastage is always my
first stop. Tigerlogic is another example (lets see where it goes).
The interesting thing is the key to success is to put any mention of
Pick "in the closet". This is very unfortunate to those of us who need
to convince the boss that people actually do rely on this technology.

That is due to ULTRADATA Corp