We are pleased to provide you with a practice for governing the Datavault model including automatic raw model generator
This Datavault generator is freely brought to you by Acceliance, feel free to reuse it and adapt it for your own needs
Please run the installation steps explained in the Github repo
Next, open Modelio, switch workspace to current Github repo and then open project
Before generating any production Model, we must set several things
The Modelio project used for illustrating Governance for Datavault is Retail domain, involving sending products into physical stores generating tickets.
The Model is present in the project as follows:
Sprints are governed at the single attribute (inside the objet) level
We will set the sprint 1 for retail Mesh/Product as the visual model diagram (previous screenshot), click the right mouse button on the tree node (as selected in the previous screenshot)
The diagram shows the exact expected output so we will use the usage cartography features of Acceliance practice:
When opening the Excel file (you will find it in the cartography folder of the project) you will notice that all the data from the diagram appears into the list
Now set Sprint1 for the whole column of Data Mesh/Product Retail
After saving and closing the Excel file, now import the usage cartography back into the model referential
When selecting any attribute in the Model diagram, you will notice the the sprint is set into the usage cartography stereotype
The Model contains all the metadata needed to generate code for actual deployment on the Data platform
Modelio uses Python language to script against the Model
Notice that the script is based on the usage cartography as seen in the <<bookmark-govsprints,Governance for sprints> part of the practice
The script is invoked from Modelio as this
Select the generate_datavault_for_sprint.py script to run
The DDL script for Snowflake is generated into a dedicated folder suffixed with the today date
The Model can be annotated with extensive documentation, as with cartography, the contents is governed using Excel (must be reinjectd back into the model)
The data modelling workshops wtih the functional experts is the best moment to capture business definition on the artefacts and it is possible to keep this information and propagate it to the enterprise ecosystem
To export the dictionary, use the following command
Then select the proper options
The documentation is automatically aligned with the deploymebnt into the Data platform
From capture
To deployment
We have seen the following items regarding Datavault Governance:
-
Viewing the Model in a fully Business & graphical manner (with no technical considerations at all)
-
Opening the opportunity to workshop/communicate with non-IT people
-
-
Governing the sprints at the level of the single data inside the object/concept
-
Using the graphical view to produce the sprint (WYSIWYG)
-
-
Using the Model metadata to generate physical implementation perfectly aligned with the Business view
-
Thus creating an automated continuum of Architecture
-
Using the UML stereotypes to control aspects of the physical generation
-
There are missing functionalitties such as:
-
Transaction Links generation (those links are used to optimize sql object volumetry, the technical pattern is that the object must have only relations of cadinality Many To One, in the case of the Retail Model, the ProductBuying object is a right candidate)
-
Multi Satellite generation (this feature can be implemeted using stereotypes and then generating several satellites for one Model object)
-
You may adapt the script for other platform such as Postgres (modify the type mapping table in the Python script)
All of the improvement can be added into the Python script
If you like our works and freely reuse them for your own projects, please give testimony on our LinkedIn company page