dFakto vous accompagne

Data Vault separates concerns to focus on speed and adaptability

Data Vaults are agile, but what does that mean in simple terms? In simple terms, it means you

Data Vaults are agile, but what does that mean in simple terms?

In simple terms, it means you don’t have to “eat the whole whale” at one time. Indeed small bites are more effective for achieving data visibility. You can even task the meal to a collection of people all at once, and the pace can increase without errors. It’s because the Data Vault is highly structured that you can split it up into pieces and (without any difficulty) join them back together again. It means you can do it all at once, in small increments, or something in between and at your own pace, according to your own resource constraints.

Just like Lego (snap it together, build out)

Your business decides the core structure of business concepts, and the data snaps onto these concepts, rather than being forced into more traditional styles of data modelling. You can start right now with as little or as much data as you have. Add to it as you figure out what your future requirements are instead of designing a full data model first. And because Data Vault is atomically structured you can use the tiniest piece in a dashboard from the first time you have that smallest bit of agile data available. If your data grows, it won’t affect what you have already measured.

You might want to do a trial on a small scale to evaluate the benefits before expanding it to the entire enterprise? No problem, Data Vault won’t waste your time meaning whatever you do now will not change allowing you to grow and expand your enterprise data warehouse once and one time only. You may choose to scale quickly by setting parallel teams each building up separate agile data stores. With Data Vault, these can all be easily synchronised through the separation of the core structure from data and can be merged together afterwards. If teams agree on a simplistic core structure of business concepts and relationships, they can each develop on top of the shared construct. It isn’t a model, it’s an agreed way of connecting business entities – something you can achieve on a whiteboard!

  • Start anywhere, evolve elsewhere, bring it all together anytime: Integrate business domains driven by real business priorities. Relations between different business domains can be established at any later point in time (doesn’t need to be thought of upfront).
  • Start small, grow any size: It doesn’t matter whether you’re building a small data warehouse for some Master Data or a full enterprise data warehouse. The advantage of Data Vault projects is that they bring results very early after you start the project, and because they are based on business concepts more than data, they are highly flexible as they grow.
  • Automate from the start for fast iterations: The key to Data Vault is deconstructed activities which are rapidly repeatable. Data Vaults and automation go hand in hand, indeed the simplicity and consistency of the approach encourages automation. Standardise your business concepts (on a whiteboard), generate the structure, automate the integration routines and then start growing your agile data warehouse.

Deconstructed data models improve automation

Once the key components of the business and relationships are understood (keys or IDs, like the customer reference number on an invoice), you simply hang data off of it, as you find it. It means that a disjointed approach to data gathering, if that’s all that is possible, doesn’t impact on the final Data Vault method, because that’s how it’s meant to work! Fast, slow, big small – it doesn’t matter. Work on different areas independently and then bring it all together, or start by attaching all of your existing data marts as sources and keep going from there, but with agility. It’s a very intuitive and flexible process for the business who can then follow the data modelling without any requirement other than they understand how the business processes work and can operate a whiteboard.

Breaking down and segregating the work in this way makes it a very repetitive process which scales well with automation. In fact, automation is highly recommended as the core structure of the Data Vault is simply meant to extract and load data ‘Satellites’ onto the lattice-work of ‘Hubs’ and ‘Links’. In fact by segregating the methodology into a network of data tables, the system almost requires automation to fulfil it’s ultimate goal of near-real-time business data for analytics. You’ll be pleased to hear that dFakto has already developed the key automation features you need to build your first Data Vault in just a few weeks!

You want to learn more about Data Vault?

Find out more from our friendly team of business and technical experts at: info@dfakto.com or +32(0)

Pin It