Data Architecture
- thierry sinassamy
Data Center
The discriminating criterion would be the obligations set out in Law 35 & 64 as well as the compliance rules set out by AM (e.g. business unit code of conduct, etc.). In so doing, the choice of data center should be in Quebec. Here are the Quebec and Non-Quebec options.
Data Fabric
The preferred type of data management system would be a data lake, as the data will generally be raw, unstructured data. Because of their ability to scale at low cost, the data lake could also be used :
for data backup and recovery;
for scientific data mining.
In order to meet the need for automated data integration between milestones and within each milestone, as well as a requirement of information access commissions (Quebec / Europe) in relation to metadata, the type of data architecture based on the data factory would be preferred.
A data factory is an architecture that promotes data integration between various data pipelines, both Cloud and On-Premise.
The advantages of this type of architecture include (but are not limited to)
use of semantics, data and metadata mining, etc.
improved customer profiling, fraud detection, preventive maintenance, etc.
promote data engineering and governance between source and destination (e.g. consumers, etc.)
more data governance safeguards put in place around access controls, guaranteeing that data is only available by segmentation or certain roles.
The type of data architecture based on data meshing would not be preferred, as this is a decentralized architecture that organizes data by business area (e.g. HR, Marketing, etc.) and also by products or sub-products. Data lakes can also be used as decentralized data repositories to create a data mesh. The data factory facilitates the transport of data from one point to another, and makes it easier to include AI/analytics capabilities.
