Differences between Azure Data Warehouse and Microsoft Parallel Datawarehouse (PDW)

Question

I'd like to know the key differences between Azure Data Warehouse and Microsoft Parallel Datawarehouse (PDW).

I read an article on Azure Data Warehouse here. Also, an article on Microsoft PDW.

Looking at features and functionalities, they look similar. However, I'd like to know key differences between them and in what circumstances, one is better than other.

Answer 1

At high level,

Parallel Datawarehouse(PDW) is onpremises version, where you have to maintain a virtual machine and IAAS(Infrastructure as a Service) hosted. You have to take care of patching the server.

Azure SQL Datawarehouse is PAAS(Platform as a Service) and you dont need to maintain Virtual machine and apply patches.

Azure SQL Datawarehouse differs in terms of features from Parallel Data Warehouse. Similarly, PDW also misses some features.

Missing features of Azure SQL Datawarehouse

Answer 2

Although there are many similarities, here are some key differences between the two (other than one is cloud and the other on-premise).

• APS data loading has a tool called dwloader.exe for high speed loads
• SQL DW loading recommendation is through polybase (external tables from Blob Storage or Azure Data Lake)
• APS Scaling is done through adding capacity.
• SQL DW Scaling is done dynamically and also offers pause to reduce cost

There are some additional internal differences such as the way Replicated tables are implemented and the fixed number of distributions in SQL DW.

As for when one is better than another, I would say the main reason would be either cloud or on-premises. Cloud is a managed instance with new features and patches regularly updated. Dynamic scale and Pause/Resume are also very nice features to reduce cost.