Am new to data vault, I understand using data vault for data that can change over time, but what about sales transactions that never change? does it make sense to vault this? I have a sales header, and multiple sales lines. if split into data vault, it will become some 40+ tables. Just not sure what the value is.
I donât know what to tell you except, you might be new to data modelling, data warehousing too.
Get involved in some data warehousing projects, what youâll find is many projects are âdata transformationâ projects, or deprecating legacy systems. Those tend to be 3NF and dimensional modelling analytics platforms. Or maybe Data Vault not done right
I have 20 years of dimensional modeling experience. More years of transactional modeling. 0 years using data vault. am in a shop now where the decision was made to use data vault. I see value for data where it changes over time or there are multiple sources for the same data (ie same hub). but when the data is transactional in nature, in our case completed sales events, the data will never change, and as of now have only 1 source. the plan is to parse this data into some 40 dv tables only to be combined back in a mart to look mostly like the data as it came from the source. so i question why to add so much complexity to what is fairly simple data. if i were to model this using dimensional techniques, it would be 1 fact and maybe 8 dimensions. also there is much volume and the data will be eventually streamed in, analytics want real time. views on the dv layer to create the consumption layer will not perform well (have already tested), so then need to materialize the consumption layer. and that adds even more complexity. guess i am trying to understand when to use dv and when not to - if that is even a question to be asked.
1 Like
In that case start here:
Transactional data or not, it does not matter to a data vault
Hello k50!
You should use a non historized link for your data that is never going to change. Google non historized link. Good luck
Hi K50!
Data vault 2.0 is a âSystem of Business Intelligence containing the necessary components to accomplish Enterprise vision in data warehousing and information deliveryâ. As such, we generally like to develop a more strategic approach that n tactical approach of a fact and 8 dims âsolutionâ. Most enterprises are more than just sales transactions, and want to use data to make intelligent decisions about how to manage the business. When you step back and look into the businessâ intention around this dataset and the questions/decision they need to make - you may find integrating other datasets becomes important (customer, product, cost, market, distribution etc.). Data Vault enables a system of data warehousing and information delivery that can be built incrementally using Agile principles to focus on business intention, rather than the number of tables you have. That is irrelevant in modem, metadata driven automation Business Intelligence systems. Focusing on adaptable, auditable, sustainable solutions rather than a series of tactical, brittle, one-off solutions is what we strive towards. If that is not a fit for your enterprise, so be it.
Also, your comment on view performance requires more justification/explanation, as there are many, many Data Vault customers currently having success with this approach - but many factors can lead to poorly implemented complex implementation tests, being new to data vault may also be a consideration.
I hope this helps you understand when to use DV and when not to, I would be happy to further clarification and communication regarding you enterprises Business Intelligence success.
Bruce