The Determination Process

The paragraphs below describe the process behind determining whether or not a Data Warehouse is the appropriate solution to meet the Client need or whether a reporting repository with specific operation requirements will be a more appropriate solution

It is important to gain a full understanding from the Client the outcome of any exercise pertaining to the storage and re-use of data for Management reporting.

Remember that the wrong decision at this point will result in a process which will most certainly exceed both the budget and deployment timetable.

Ask the customer to describe the outcome they are looking for in relation to the system. This may take the form of a presentation of the information they already are using or potentially a demonstration of an outcome made available by another party.

Does each outcome require input from various sources of data. A simple set of examples would be:

• A report on how many Term Deposits with balances over $1,000 are held in a certain location.
• A report on how many saving bank deposits are made every day by account holders in a certain branch.
• A report on how many 15 Year olds hold bank accounts with the institution.

Each of the examples above requires data primarily from one source.

• The first requires information purely from the Term deposits database.
• The second requires information from the savings bank database.
• The third requires information from the customer database.

All this information is valuable business information for marketing but can be answered from a simple reporting repository or indeed the operational database in which the information is collected.

Examples where a data warehouse would be of value using the same information would be:

• How Many 15 year olds with accounts in the bank and term deposits over $1,000 are likely to make savings banks deposits next month.

This request would take potentially many hours of processing against multiple sources of data, tying up valuable operational computing resources and need additional interim storage in order to answer a question that may be posed only once.

In this case, a properly organised data warehouse with a method of storage designed to multi-thread queries against more than one database at at a time would be far better suited.

I am only quoting simple examples here if you multiply this question by 1,000 requests per month and you can see the benefit for Management and Marketing this may give.

Engagement with the client to understand where they see the primary usage of this data will assist in the first decision you are asking them to make.

Do you really need a data warehouse to do this?

Or

Will a reporting repository that capture the necessary data from each operational database but keeps them separate be the right answer?

Or

Can you build a reporting solution that runs directly on the Operational systems be acceptable due to low system loads etc?