Designing data-centric solutions in the cloud can be a challenge. Not only do we have the standard challenge of meeting customer requirements, but we also have to deal with an ever evolving data landscape of tools, paradigms and platforms. In the good old days of data warehousing for example, we had relational stores, often coupled with flat files/rogue Access databases etc, which were then fed nicely into another database using an ETL tool and then made reporting ready, often following one of the prevalent design methodologies for these solutions (Kimball, Inmon, Data Vault etc).
Fast forward to now (2018), and whilst the data warehouse is far from dead, the modern data platform is a much evolved beast featuring fast data, slow data, big data, small data, cold data, funny-shaped data, not-funny – you get the picture. All of it is needed to be processed in a way that can derive actionable insights and intelligence for us poor humans.
To help with this, Microsoft have released the Azure Data Architecture Guide, and it provides, to quote directly –
…a structured approach for designing data-centric solutions on Microsoft Azure. It is based on proven practices derived from customer engagements.
The guide’s main focus is the difference between relational and non-relational data, the design paradigms of both and the best practices, challenges and processes for designing these kinds of solutions.
It is by no means a deep-dive in the nuances of each component, but it gives good, broad content in this subject, allowing us to understand and appreciate the core concepts before diving into specific focus areas as required.
And a PDF can be downloaded here.