Hi All,
I've received a few off list responses to my comments on IWS......
I thought it appropriate to put in writing some of the facts and my opinions about this.
I'll do it in bullets to make any discussion easier. Here are some comments I am getting and here are my answers.
1. More information please.
A number of people have asked me to provide more information on IWS. I can't provide much more than I have provided.
I am not a Sybase employee, however, I spend all my time implementing IWS around the world. (Well, not Australia.) I've signed all the appropriate non-disclosures.
If anyone would like more information on IWS go to
Also, one of the listers is Jon Simmons (Jonathan.Simmons@sybase.com ) and he has made the point a couple of times about IWS and IQ. (However, we all know that hearing how great a product is from the vendor is very different from hearing how great a product is by someone who has no particular axe to grind in that direction.)
Jons a great guy, very knowledgeable, and a good friend of mine.
Everyone on the list should feel free to ask Jon about IWS. Obviously, he does get paid for selling IWS and so if you ask him questions you must expect that he will want to know if you are in the 'free education'
category or the 'seriously interested' category.
I'm sure he would be happy to answer general questions about IWS to the list. As an employee, he can probably say more than I can.
------------------------------------------------------------------------
------
2. Why talk about something and then not explain the details?
Well, this list seems to have evolved into devoting most of it's discussion to modelling. (It wasn't always that way.) And many of the modelling questions put forward are elegantly handled in IWS. When I see these questions I have two choices.
A) Let people on this list know that IWS handles the situation even more elegantly than the explanations offered or
B) Say nothing and leave the list non-the wiser.
I choose A) because I believe it is of more value for people on this list to know there is another alternative, even if they choose not to buy it, than to leave them believing there is no better alternative.
Now, if the listers on here vote with their reply button and ask me to not mention such things, that's fine. I've got lots of other things to do with my time..;-)
--------
3. Why are you 'touting' IWS?
Firstly, 'touting' implies selling and being paid for selling. I've said here many times, and I'll say it again, I don't get paid for selling IWS. I do get paid for implementing it. I have (very) occasionally been paid for my time to go to a prospect and explain IWS in detail as a person who does not get paid for selling it. I have written many papers for Sybase prospects on why IWS is 'a better mousetrap' and none have been 'paid' work. I have even taken calls from prospective IWS customers who wanted to know more details about the product and check the claims that Sybase were making.
I talk about IWS nowadays for the same reason I started talking about Star Schemas in 1993.
Both are a great idea whos time has come. (Notice it was a good 6-7 years later that star schemas really became 'mainstream'.) I think maybe this time it will be faster because IWS is 'evolutionary' while dimensional modelling was 'revolutionary' in most IT shops.
(By the way...just on touting to this list. The people on this list don't 'buy stuff' as far as I am aware. (Maybe datamirror has managed to sell some software to people on the list...;-) As a contributor to this list for over 5 years now I believe the only income I could say came from the list was Ralphs kind request to write a column for him once.
Not much of a return for what must now be hundreds of hours of posting appends to this list...)
4. What's you opinion about IWS? Really, off the list, you can tell me, I won't tell anyone else.
Well, it's the same off the list as on the list. I think the modelling techniques are excellent. In the 3.5 years I've been implementing it I have only been able to come up with 1 innovation. And even the original designer was impressed with that one......To use a model for 3.5 years, day in, day out, and not be able to find a 'second innovation' says a lot about the modelling techniques used.
The major innovation of IWS is that it can store archival data inside a dimensional model in a non-lossy format without loss of flexibility or performance. Quite amazing.
It can also achieve performance of update that will satisfy most ODS requirements.
By doing away with a separate archive layer the overall project costs around 30-40% less than if a separate archive layer was built. So IWS is going to be cheaper than building separate archive/analytical layer and probably a more than just designing and building your own dimensional DW.
-----------------------------------------
5. So IWS is not CIF then?
I really don't get why I get asked this question....it makes no sense to me....
IWS is CIF just like many other things can be components of CIF. There is a 'logical' archive and there is a 'logical' analytical layer, and there are 'logical' data marts. It's just that they are all in one physical database. IWS implements time variance and stability analysis inside a dimensional model with performance and simplicity.
---------------------------------------
6. But IWS only runs on IQ.
I don't know who started this rumour.
It runs on anything that is dimensionally aware. And that's all the major databases. There is even a copy running on Teradata, but it was heavily updated to allow reasonable performance on the NCR machine.
By the way, I like IQ as well. Given that most customers want the DW to be 'cheaper and faster'. IQ will halve the cost of the HW/SW portion of a large DW and run at least 10x faster than it's oracle equivalent. IQ meets the 'cheaper and faster' criteria quite nicely. But, lots of people buy Oracle for the DW database. Fact of life.
7. How much does it cost?
Call Sybase. In my opinion....cheap..;-)
So, that's about it. There's a bit of history at the bottom for those that are interested.....
Summary.
Some people will buy IWS, some people will buy IBM or NCRs DW models, some people will read Bill/Claudias/Ralphs books and choose to build their own in the way that they please......the reasons people will give for whatever they do will vary.
All are valid and valuable ways to move forward. No-one has the 'one true mousetrap' when it comes to building DWs.
My contribution here is merely to point out to people in the list that my opinion is that IWS has some great modelling techniques imbedded in it. Those that do want to learn about them know how they can find out.
I feel this is better than not sharing with the group at all....
And that's about the last I'll be saying on the subject I would think......
Best Regards
Peter Nolan
Data Warehousing Consultant
Mobile: +353 879 581 732
Homepage:
Background 3A.
Some background just in case you are interested.
In 1991 I developed some 'new' (I had never seen them before) modelling techniques to improve performance of what were then BIG DWs. 5-10GBs, sometimes even more!!! These techniques improved performance 100-300 times over 3NF models. I spoke at many IBM conferences and visited many IBM customers explaining these techniques. They were a hit. Many large companies in Australia adopted them. One IBM customer credited me with enabling them to do things that would have cost USD12M in IBM hardware to do the way they had been going, so they had never done it. Many large companies said these techniques made the 'impossible' 'possible'.
In 1993 I got the huge opportunity to go to the Metaphor users conference and talk face to face with many household name companies who used star schemas.....once they told me what they were able to achieve and how they achieved it I realised the dimensional model made what I had been doing redundant. I became very 'pro' dimensional modelling, even though it killed what I invented previously.
I went around all the same IBM customers telling them about star schemas. I was expecting a similar positive result as 2 years earlier....but, strangely enough, I was shown the door more often than not for talking about something so radical. Truly, shown the door!!! By the very same people who were using my previous, now redundant, ideas!!!
I didn't get it at all.
I was ever so grateful to Ralph and co for writing a book because I could point to the book and say 'this is real' people have been doing this for 10 years. 400+ companies have them!!! You can't just kick me out like you did 12 months ago!!! Heck, for 2 years I gave away a copy of Ralphs book with just about every presentation I did along with the phone number of where to order more copies!!!
So, here we are again. IWS is to star schemas what star schemas were to what I was doing in 1993.
Again, the knowledge is proprietary. It must be paid for.
Just like in 1993, when I tried to learn about how to design star schemas and how to populate them I was told 'no'. Despite Metaphor being owned by IBM and me being an IBM employee.
The deal was buy from the Metaphor Consulting Group, or not at all.
So, we had to buy consulting from the Metaphor Consulting Group. It was very, very expensive so we only bought a little bit of time.....then we spent 18 months figuring out the rest ourselves because that's all the customer could afford!!!
So, there is strong precedent for not disclosing proprietary information about models.
And I would also point out, no-one on this list is talking about IBM or NCRs models at all....