Warehousing Whys and Wherefores - International Consultants' Guide, May 2001
There is no argument that it would require a very brave individual or company to state that a business can survive without information.
Equally few would say that information is achieved without data management and efficient delivery of data. The delivery of information occurs when data is presented to business users in understandable, legible and visual format.
We are told the world is in the Information Age. Why is it, then, that the term 'data warehousing' can provoke such consternation amongst business and IT professionals
Projects have been stopped because someone saw or heard the term data warehouse or, more recently, enterprise data mart.
Etiquette also often forces generalisations when discussing data management, with terms like information cubes, reporting repository, analytical and operational data stores, samples and performance-tuned databases.
Why is it that an approach to building the infrastructure for a business solution generates such a need for jargon and such strong emotions, for and against?
A number of leading exponents in the data management arena, whom one would expect to be cohesive and coherent on the topic, are polarised in their proclamations on data warehousing. Argument abound around success rations and measurements, the critics claiming that a new approach, structure or technology is needed to reverse the trend. Interestingly, many have just such a product! The more positive exponents claim that measuring success based on past performance is neither valid nor relevant. They argue that, as the industry matures, improvements from experience recognise and resolve the shortfalls of the past. Where does this leave the business decision maker?
Confused, apprehensive, distracted and frustrated are all feelings that business decision makers experience from this maelstrom. They seek guidance and assurance from their IT people. However, if the IT people are not achieving success from data warehousing today, how can they reassure the business? This is where the value and benefits of a delivery-focused IT consultancy are best applied. The IT consultancy can be instrumental in working with IT in moving the decision makers away from the theory of information delivery into the practicalities of business solutions using the business data for information purposes.
Today more than ever, information differentiates the successful from the also-rans. With the recent explosion of the world wide web and the 'e' systems that now constitute business channels to market, suppliers and vested parties are much more greatly dependent on information. The web race has altered the competitive structure. The sprinters are being overtaken by the middle and long-distance competitors who are seeking real revenues and profits.
The longer business race for survival will be won by the business that can personalise the passive, if not cold face of technology. Information is the key to personalisation of e-channels. Data management and delivery of the data is how information will be brought to the channels. Therefore data warehousing is becoming increasingly critical to business success. The key is to move beyond theory into delivery.
To Delivery and Beyond
Much depends on the business strategy and IT implementation maturity when assessing where, when, how and what will benefit from data warehousing. Moving to 'real-time personalisation' of channels when the integration of systems is not available is unlikely to be successful without significant investment, commitment and vision.
Even with all these elements in place, the risk of failure is great as the project duration is often beyond the tolerances of the stakeholders and other vested parties. However, leveraging a small number of disparate databases to enhance management information or primary channels performance is significantly more likely to succeed. In addition, this partitioning or phasing enables the business to evolve into an information-led organisation.
Does partitioning or phasing imply that the enterprise data mart is the most suitable architecture available? The answer is not that simple. The construction of data warehousing should not be bound by a particular data structure, model or methodology. The business needs to hedge between optimal and minimal implementation factors. The factors for the hedging should be the businesses objectives and goals, the priorities and fiscal constraints and preferred method of deployment.
Setting an Example
Take as an example the analysis of a proposed data warehouse project for a medium sized organisation. The organisations business strategy is to partition its business functions into disparate business entities, with the aim of floating each entity.
The current IT systems status is that a number of independent systems exist at different locations. Two of the key business functions share a common platform and database product. The two business functions are also connected by the corporate WAN.
The remaining business functions have an assorted mixture of platforms and databases. Those systems partially connected to the corporate WAN, though most operate on LANs with no corporate connectivity.
The business clients can be common amongst the business functions and undertake similar activities, though these activities are performed in a different manner. The question is how does this business move forward with data warehousing?
Did the business select a three tier architecture? The answer is no. Nor did the organisation decide on a enterprise data mart. Neither approach would have supported a ready separation of the business functions into independent business entities.
The choice made was to generate independent data marts for the two business functions that had relatively easy connectivity. A leading ETL toolset was used to populate the two data marts and enhance each with selected external data.
The data marts serve the business functions in different manners. The first business function uses the data mart to support its call centre in evaluating a clients potential lifetime value supporting client queries and providing account details.
The second function uses its data mart to provide "real time" management information on client activity and web infrastructure performance. This business function also leverages historical details, including those from the call centre system, to analyse the impact of the web interface on client activities. The longer term aim is to bring the information contained within the data mart into the web interface to provide client specific customisation.
In the example provided, the challenge is to not to identify the right architecture, implementation, approach or database schema but to determine whether the solution provides the information the business needs and whether it will continue to fulfil this service to the business in the future.
The recent development of portals is further extending these challenges of information sharing on e-channels.
Portals operate on the premise that by enlarging the business capacity for providing an information channel to a broad range of users, internal and external, the business will benefit.
The challenge here is to package the information to be meaningful to this broad range of users. Meeting that challenge will push data warehousing into new areas of data management and the delivery of data. The guidance of the theorists will be needed as much as it has been in the past.
One stream of argument surrounding the portal challenge is that this is bringing the management of the business information into the internet paradigm. In a world where, primarily, we have been accustomed to some sort cascading model of responsibility and authority, the internet is anomaly. The internet was founded on the principle that there are no critical hubs. If a hub is lost or removed, the network enables the system to remain operational if not fully in tact. This structure best supports an information strategy where localised data hubs exist that are accessible to all though maintained by a few. The challenge is, how do you maintain and validate data content?
Obviously the theorists will assess and debate the pros and cons of a variety of approaches. From these debates some very valuable and practical solutions will be identified. However, ultimately it will be the IT implementation practitioners who must recognise and resolve the issues the portal world throws up.
Value of Experience
OCS Consulting started delivering a range of "single view" repository based solutions in 1986 when it developed its first "customer centric" solution. Also during this period the term data warehousing, and numerous implementation methodologies and approaches, became common. OCS assessed, applied and adopted a number of the methodologies and approaches.
Hence it has experience in a number of data schemas, including the poplar of populist architectures such as ROLAP, MOLAP, HOLAP, Star, Snowflake and many more. Data structures and models, it has found, must be considered according to the clients needs, systems, experiences, skills and preferences. Solutions must meet the business objectives, whether performance, consistency, completeness, budgetary or technology rather than being led or leading by fashion or popular theory.
This pragmatic approach will not always go down well with software vendors and evangelistic data practitioners. However, the approach does reap rewards of high client satisfaction, successful business solution, implementations and motivated teams.
It seems that data warehousing is not only live but rapidly growing, evolving and extending as technology empowers businesses with the capacity to broaden their audiences for information. The world of "e" is opening a new and exciting phase of data warehousing.