This article originally appeared on the BeyeNETWORK.
In the past – in the day of OLTP systems – metadata was centralized because the systems that were running in that day and age were also centralized. Makes sense.
But something else was going on. For the most part, the systems that were built in the early days were operational, transaction oriented systems. These systems entailed very structured transactions, where the processing was rigidly enforced by programmers' code, such as COBOL or PL/I. In these early systems, there simply was not anything such as ad hoc processing or end user reporting.
The same transaction was run repeatedly. An operator using these systems might use the system 100 to 500 times a day, depending on the job of the user.
Let’s now consider the role of metadata to the users of repetitive transaction processing systems. How often does the end user – the operator – need to use metadata that is found on the screens that govern the entry of the transaction? In reality, after an end user has executed the same transaction many times, the end user pays no attention to the metadata.
To use an analogy, when you drive to and from work each day, as you have done for many years, how much attention do you pay to the road signs? The answer is very little if any. You already know where the stop sign is. You already know where your intersection is. You drive the roads every day and the road signs just are not useful to you. You already know them by heart.
But the computer profession did not stop with centralized mainframe systems. Today we have personal computers, workstations and end user reporting tools. We have business intelligence which is designed for end user manipulation of data. We have people who are first and foremost businesspeople using computer technology.
The nature of their business is exploration and analysis. There is very little repetition of transactions among this community. These people are out of the box thinkers, and many of them never execute the same analysis twice. Metadata plays a special role for these people. Metadata allows the end users the ability to find out what has been done in the past, so past analysis can be reused, rather than having to be redone. And metadata allows the end user analyst to see if the end user analyst is “fishing in the right pond” before an analysis is done. In a word, metadata acts as a guiding force for the decision support analyst.
In its own way, metadata is a road sign to the end user analyst, in much the same way that metadata was a road sign to the transaction processing end user. But there is a very real difference in the way these two communities view road signs. An end user analyst is someone who seldom goes over the same road twice. Road signs are very important to the end user when going over new territory.
Consider a trip from Wisconsin to Santa Fe, New Mexico. How important are road signs to you? If you have never before been to Santa Fe, road signs are vital. You may be heading west on I-70 when you should have turned south in Omaha. If you want to arrive in Santa Fe in a timely and orderly manner, then you pay attention to road signs. All the time.
So it is seen that there are different roles that metadata plays to different communities.