Who, how, what - why bitemporal data?

Who, how, what - why bitemporal data? Who does not ask will not be wise!

Sci-fi fans will know that a new actor playing Dr Who, in the long running TV show, comes far more frequently than major changes to the data technology landscape.

When actors are recruited prior to regeneration, I doubt the prospective Time Lords have to define time before they ever read a line about dealing with dangerous Daleks or calculating Cybermen.

But that is the challenge facing many data teams trying to create the best data platform solution – in the Cloud if not outer space – what kind of time is the business dealing with and what are the businesses’ requirements for managing complex data storage and data analytics?

That was the question posed at the May meeting of the Data Vault User Group when Tedamoh’s Dirk Lerner, presented ‘Who, How, What – Why Bi-temporal Data?

It was backed by the rider that: ‘Who does not ask will not be wise!’

Having explained, for those who are new to the concepts what the scope and limitations of databases are today, whether they are temporal (current), bi-temporal (with two timelines, transaction and business time), and tri-temporal (transaction, business and decision time).

Dirk’s timeline then visited the classic question facing every business using data analytics – what did you know and when did you know?

The reality for many organisations is that data can arrive out of sequence, and there is the need to store the “reality” in past, present and future formats, making corrections against the past as the picture changes.

The ability to query through time and analyse decisions, let alone any legal regulations or audit requirements a business faces, complicate matters further.

Dirk explained how the different kinds of databases could cope or deal with such scenarios.

He also explored the likely probability of a data error in terms of external (data vendor information) or internal factors – the latter could be down to human or machine error via an automation process.

In a bi-temporal data model, the business can make corrections, while keeping the errors recorded, allowing for audit access and fully traceable data, which breeds confidence in any changes made.

When it comes to decision analysis, the bi-temporal model can deal with the late arrival of data which can be correctly historicised on the timeline.

The process updates the previous historicised data so it is the correct data – again increasing confidence in changes.

Watch Dirk’s full presentation above.