Data Warehousing
All systems can be characterised in three elements: input, process, and output. The process is turing raw data into information.
However, for information to be usable it has to be accurate, and accessible. Accessible also means timely; the key management indicator that arrives a week late on the chief executive officer's desk has lost significant worth. Therefore there is a hidden cost associated with "slow" information which must be factored into any calculation of the potential return on investment in a data warehouse.
If the source data is currently held in legacy systems, there will be a cost in extracting and organising that information. But your organisation will face that cost any time the data needs to be extracted. It is possible to extract the data directly, but that involves a high cost to gather the information every time it is used. Extracting legacy information into a data warehouse allows all kinds of analysis to be executed cheaply and repetitively compared to extracting from sources on each occasion.
The previous article on Data Quality outlined many of the challenges associated with extracting from source. At this stage, let us assume that our input data is of good quality. But in a sense, it is still raw. To use the industry acronym "ETL", we have Extracted the data from its sources, but as yet we have only partially Transformed it. The data may be cleansed – now it needs to be organised.
Latency and Performance
Information consumers will always want their information as quickly as possible. When dealing with large data sets, latency is inevitable. One way of addressing latency is better hardware, faster processors, quicker disks, more and faster memory. There is definitely a place for this strategy, but there comes a point when more hardware does not mean better performance. Spending more money on hardware at this point does not reap a return.
With software such as SAS, there is the question of configuration. Even where SAS components are distributed across many servers, it is still possible that SAS will experience contention where little or none was anticipated. Configuration is such a complex issue, it is one reason why many companies ask Amadeus to run an Efficiency Review.
Design
So, assuming that the data is of good quality, and that the hardware environment is optimal, and ditto the SAS configuration, what of the data warehouse itself? Designing a data warehouse requires a knowledgeable Consultant who can translate the business rules governing data relationships into a warehouse structure that promotes efficient retrieval of information.
The Ability to Repeat
The cost of a data warehouse is likely to be directly correlational to the cost of extracting and disseminating the information from legacy systems once. The benefit of a data warehouse is that having undergone the exercise once, the cost of analysis and reporting on that information drops by a huge amount: pounds become pence, dollars and euros become cents. Because the cost of analysis becomes that much cheaper, it also means data is used and reused many more times over. Timely information translates into better returns on investment of time throughout the organisation.
Amadeus have successfully delivered several such projects before. If you would benefit from our Consultants reviewing your data processes, data warehouse or if you are planning a new project, call us on +44 (0)1993 848010 or contact us via email by clicking here.
