Classic data warehouses are relational databases whose structure is optimised for fast querying of the contained data. The power of data warehousing lies in the effective linking of data. This creates new insights into the underlying business. Typical applications are the combination of financial data with that of production by, for example, the profitability of products per customer, per shift, per lot size, etc..
Structurally, “stars” or “snowflakes” are usual patterns, with a fact table surrounded by one (star) or multiple (snowflake) layer(s) of dimensional tables. Such a data structure creates cubes within a space of an arbitrary number of dimensions. A space of n dimensions has n*(n-1)/2 axes of rotation. This is the number of Excel sheets that can be extracted from a data cube to incorporate the ever fluid dimensions of said cube when projecting to the two available axes of Excel. The data cells are aggregated over all other dimensions – in the simplest case these are sums.
A data warehouse can therefore enable evaluations from the existing data in seconds, which would otherwise take a very long time – for both computers or people.
As is so often the case, the art of data warehousing lies in the balance between effort and benefit: databases that are too large can become slow, require strong hardware and a lot of maintenance. Undersized or poorly structured warehouses have trouble – or are indeed completely incapable of – answering questions.
If you have data from which you would like to learn, please contact me. I will analyse your existing data, structure it into a warehouse and implement its management. I will also support it in both operation and with training requirements.