BLOG

Modern Data Warehouse with Azure Analysis Services

28-08-2019  0 Comment(s)

 Microsoft Azure and all its PaaS components such as Azure Analysis Services, I was routinely sticking to Microsoft’s on-premises BI stack. With this constrained view on technologies, also some restrictions in terms of cloud-based technology re-interpretations came along.

 imagine the following situation: you have your SQL Server Data Warehouse on-premises with regular relational tables and you are using a Tabular Model on top of it. You are building fancy dashboards and hosting them on your Power BI Report Server or you have them running in the PowerBI service. Doesn’t sound new to you, right!?

Now you can do this sort of stuff in Azure. It’s the Modern Data Warehouse approach (for more details on MDWH check this out). “I know!” …most would say by now. However, what I have experienced in most customer contexts, is that people are aware of this rigid, classical approach using a relational component, either Azure SQL (managed instance) or Azure SQL Data Warehouse, on top of their data lakes with subsequently attaching an Azure Analysis Services Instance hosting some Tabular Models on top of it.

There are many good reasons to use relational components, such as SQL DW for instance, especially for big workloads, considering scalability, reliability, as well as available skillsets in the SQL context. Another reason is to have an infrastructure that helps you delegate tasks to dedicated teams in the context of ETL or data reliability and consistency. I really do not want to understate the importance of one accurate and trustworthy source of truth for BI solutions.

On the other hand, we may also face scenarios where we do not need to store data historically (SCD) or where our master data gets fully loaded each day according to given prerequisites by source systems. A big data context could also be legit, where we may not have the need for a classical data modeling approach. Maybe we want to visualize a massive amount of data. For such scenarios, a relational component between data lake and Azure Analysis Services may be causing overhead.

Comment Here

Comments

No Comments to Show

WE ALWAYS WORK WITH :