Bill Gates started Harvard and did not finish. Therefore Bill Gates is a failure. Or so a survey from the Data Warehouse Network would have us believe.
Does any of this make sense? Bill Gates is the wealthiest and most influential man in technology in our century and anyone that thinks Bill Gates is a failure is just plain stupid. Bill Gates is a stunning success and everyone knows it, except perhaps a survey from the Data Warehouse Network.
An survey from the Data Warehouse Network has been taken and has used a definition for failure of data warehousing which is worse than circumspect. The survey has defined a data warehouse failure as occurring when we have to "abandon the initial design, scope, strategy, infrastructure or technology" of a warehouse design. If we used this criteria for failure, in the case of Bill Gates (who admittedly is not a data warehouse) - that Bill Gates started Harvard and did not finish - then we would have to agree with the survey that Bill Gates is a failure.
If failure is defined to be sticking blindly and dogmatically with a vision or goal made at a some previous point in time regardless of the current facts and circumstances now facing an organization, then all sorts of strange proclamations can be made.
What is wrong here is not that Bill Gates is a failure (he isn't!) but the definition used in the survey made by the Data Warehouse Network. Does anyone believe that a failure is someone or something that must fulfill all life's goals even when those goals were made under a different set of conditions at a previous point in time? Of course not. That simply is not the way life works.
The problem is that the survey from the Data Warehouse Network has made up a definition for data warehouse failure that no more fits a data warehouse than dancing shoes fit an elephant. The Data Warehouse Network will tell you that 70% of data warehouses are failures because of the results of the survey. The survey will tell you that a data warehouse failure is construed to be a data warehouse where a data warehouse designer has changed their initial design.
This definition of a failure is simply grossly out of phase with the way that data warehouses are built. People don't fail when they learn enough about data warehouses to alter their initial design. In fact that is how little data warehouses grow up to become successful large data warehouses.
And amazingly in the next sentence, the Data Warehouse Network tell you that in most cases where the data warehouses "fail", the data warehouse design is altered and people carry on with the data warehouse development. If data warehouses were really a failure, then why do people continue to build them after they have "failed"?
Never mind that in the best books and methodologies on the design of data warehouses that it is stated that the nature of data warehousing is to do iterative development where designs, strategies, technologies and the like are regularly restated. Especially initial strategies. When people alter their initial design efforts for a data warehouse, they are following the time honored path of iterative development. Simply stated the survey from the Data Warehouse Network is calling a data warehouse failure something that is anything but a failure.
If the survey from the Data Warehouse Network had really wanted to have measured data warehouse failures, they might have used some criteria that looked like:
-
look at organizations that have gone through multiple design and development iterations, not just the initial iteration,
-
look at organizations that understand the differences between a data warehouse and a data mart,
-
look at organizations that have or have had a fully populated data warehouse,
-
look at organizations who have had a full community of users examine and try to use the warehouse and who have then abandoned it, and finally
-
look at organizations who, after having done all of the above, have not restarted the warehouse effort.
Anyone who qualifies under these criteria is a data warehouse failure. The criteria used by the Data Warehouse Network is grossly misleading and is plain incorrect.
But the survey conducted by the Data Warehouse Network is not the only misleading source of information the industry. A person that is misleading the industry is Doug Hackney, "Mr. Data Mart" himself. Doug Hackney regularly and loudly proclaims data warehousing to be a failure. Doug publicly and often quotes a failure rate of 70% for data warehousing. In fact Doug has so often quoted the rate that the number has entered the status of becoming conventional wisdom. Doug has even been quoted in Forbes magazine talking about data warehouse failures.
The message has come to be - "every one knows that data warehouses have a failure rate of 70%". Or do they really know that?
What about the truth? What about the reality of the numbers stated by Hackney? When asked about the 70% data warehouse failure rate, Doug cited three sources:
-
the survey from the Data Warehouse Network,
-
an Organization and Technology Research (OTR) study, and
-
Price Waterhouse.
The survey from the Data Warehouse Network - as we have seen - is seriously flawed.
Mike Schroeck, a partner at PW and a spokesman for their data warehouse practice has stated that he absolutely does not support Doug Hackney's numbers. In fact, Mike says that he has not seen even a single failure of data warehousing. According to Mike (and we quote verbatim) - "I don't know where Doug gets those numbers from, certainly not us".
Now let's go into the OTR study, copies of which Doug was happy to forward. The OTR study surveyed 1500 sites in Europe as to their data warehouse activities and experiences. The results of the survey were announced in early 1997.
The first problem with the OTR report is the timing of the report. In late 1996 or early 1997, this survey was made against a very immature data warehouse audience. There are not today 1500 mature data warehouse sites in Europe. How in the world could there have been 1500 mature data warehouse sites in early 1997?
The conclusions of the study were:
To draw the conclusion that data warehousing was a failure because companies could not find financial benefit is a real stretch. Companies do not find financial benefit until data warehouse technologies and their implementations are complete. There certainly were not 1500 mature data warehouses in late 1996 or early 1997 in Europe. So why do we take seriously a study that purports to measure the end result of data warehousing when the majority of the organizations surveyed have no real experience with a data warehouse? Asking an immature data warehouse market to respond to the issue of long term financial payback is like asking a school of 12 year olds about the variety and the richness of their love life. Immature young people simply are not in a position to respond with any authority about things that they only later in their lives will know about.
The second problem with the OTR study is that it does not ever mention a failure rate of 70%. That number appears to be a convenient fiction created by Doug Hackney.
A third problem with the OTR study is that it apparently does not know the difference between a data warehouse and a data mart. Making a study where data warehouses and data marts are mixed together is like making a study of vehicles where some of the vehicles are Sherman tanks, some are Porsches, some are mopeds, and some are Suvans. Any generalizations made as the result of such a study are very questionable. The first thing that an organization making studies needs to do is to carefully understand the subject matter being analyzed. OTR is not sophisticated enough to know even the basic differences between a data warehouse and a data mart.
But the study does say some very interesting things. The OTR study says that - "31% of UK users currently have a data warehouse or are building a data warehouse and that figure will climb to 39% by the end of the year." Why doesn't Doug Hackney ever mention this?
Or the OTR study says that "users intend to quadruple the size of their data warehouses this year from an average of 20 gigabytes to 80 gigabytes and double the number of end users who access it from an average of 115 to 322." And where does Doug Hackney ever mention these results?
And if data warehousing is such a failure why do two thirds of the companies surveyed say that "such systems are either strategic or very strategic to their organization".
It seems that Doug Hackney is very selective about presenting what he wants, not what the facts really are.
The OTR study - if it can be believed at all - says something very different than what Doug Hackney portrays. The study never addresses failure rates (it addresses failure to find a financial return on immature warehouses) and never states that there was a 70% failure rate. Instead, the OTR study states very positive things about the state of data warehousing in Europe. But the positive facts never seem to come out of Doug's mouth.
Something is very wrong here. Doug Hackney has quoted three sources - a survey from the Data Warehouse Network, OTR and Price Waterhouse - and all three sources absolutely do not support the message Doug Hackney is giving when put up to close scrutiny. Admittedly the survey from the Data Warehouse Network supports the numbers Doug Hackney gives out but the survey is a seriously flawed product which when examined closely cannot be believed or taken seriously.
There is nothing wrong with healthy debate on any topic. Data warehousing is no exception. But the first element of health is truth, and somehow Doug has missed the truth in his presentations. A debate based on misinformation - deliberate or otherwise - is not a healthy debate.
Just as Bill Gates is a resounding success, data warehouses certainly do not have a failure rate of anywhere near 70%. And that's the plain truth.
Sorry about having to interject the truth, Doug. But the world of data warehousing deserves something better than what you have been feeding them.
凡是有该标志的文章，都是该blog博主Caoer（草儿）原创，凡是索引、收藏
、转载请注明来处和原文作者。非常感谢。