Business Intelligence Network - The Vision for BI and Beyond

I have been involved with large-scale data processing for about thirty years. Even my occasional use of the term data processing dates my experience. I must admit that, in those early days, I was attracted to the flashing lights and spinning tape drives. Amid the roar of air conditioning, it was just cool to roam the raised floors and look into the blue cabinets. This was the age of real mainframe computers …when computers were computers and could not be mistaken for anything else!

I remember the days when client/server systems were introduced based on a mini-computer server that would fit into a single cabinet. And then, I remember the days when personal computers were silently smuggled into the office late at night. The term enterprise system was born to distinguish the real computer from all these frauds.
The enterprise system was the one system (not plural) that supported the primary data processing needs of the entire company. The implication was that the enterprise systems of old had the following dimensions:

• Enterprise-Scope: Its responsibility was to support the entire company, which in those early days had the narrow scope of financial applications, such as general ledger, accounts receivable/payable and payroll.
  
  
• Large-Scale: It was big, both physically large and functional comprehensive. The assumption was that capability came through economics of scale. One large system would always be better (faster and cheaper) than many small systems.
  
  
• Reliability: Paranoia for keeping the system working properly soon became a critical dimension to the management and architecture of enterprise systems. It could never have an unplanned outage. Payroll would occur on time. And, new orders could be processed at any time.

We now live in a new age of information technology. This became apparent to me during a study of data warehouse appliances with Colin White. Having proved that data warehouse appliances have been successful with data marts for point solutions, a major issue was whether data warehouse appliances were ready to support enterprise data warehousing. In response, Colin and I focused on how the requirements of mixed workloads varied from data marts to enterprise data warehouses.

As I reevaluate the nature of enterprise data warehousing, the following dimensions for enterprise systems of the future emerge:

• Cross-Functional: Enterprise systems of old were a contradictory collection of silo applications. Today, the scope of enterprise systems is redefined with new requirements that cut across traditional functional boundaries. The quest is to integrate business processes and business information so that customers and other key stakeholders view the business from their perspective. This perspective is more than satisfying a single version of the truth. It requires multiple views of the business from the outside in, rather than a single official view from the inside out.
  
  
• Extreme-Capacity: Large-scale is now redefined with new metrics for speeds-and-feeds, rather than pounds of iron. A classic metric has been data size. It was a bragging point to say that you supported a 2TB database. Now, data size is becoming meaningless with the focus shifting to the rates of storing, analyzing and distributing information to a community of users, both inside and outside traditional corporate boundaries.
  
  
• Diverse User Base: Enterprise systems of old were reporting machines with one class of users – depersonalized batch reports. Data centers were factories that produced paper. Trees entered one side of the data center and emerged as stacks of green-bar paper. Users were reduced to locations for depositing these stacks on certain time schedules. Today, users of enterprise systems come in all shapes and colors, both internal and external to the business.
  
  
• Complexity: Enterprise systems of old were much simpler and lived within a controlled environment, both physically and informationally. Today, systems are not more of the same, but have become a diversity of loosely coupled pieces, all of which are in constant reconfiguration. Today, data is not static, sitting on large disks waiting for historical analysis. Data is in constant motion, streaming from an infinite pool. It increasingly seems foolish to catch this torrent in a bucket called the enterprise data warehouse.
  
  
• Mission-Critical: Enterprise systems of old were held to high levels of reliability because, if reports were not generated, people got upset. Today, the situation is much more critical. If the enterprise system malfunctions today, the business stops. Having people upset is the least of the problems.

What is an enterprise system in this new age of information processing? This term may be increasingly difficult to define. As technology shifts to on-demand computing, the location where processing will be performed may not be distinctive. As data external to the business becomes more important, it will increasingly reside in unknown data collections.

We will be challenged to distinguish between the enterprise system and the various components and services that support enterprise computing. It may be that the term enterprise system will be a quaint phrase shared among gray-haired folks reminiscing about days gone by.

If you found this article helpful and would like to receive the latest insights each month from the experts featured on the Business Intelligence Network, please subscribe to the UK Business Intelligence Network Newsletter.