Microsoft’s Next Generation Federated Data Warehouse Platform

Figure 1. Microsoft Federated Enterprise Data Warehouse

Microsoft’s new Hub-and-Spoke approach to data warehousing includes multiple shared-nothing nodes (self-contained database servers), with ”hub” as a centralized data model for all subject areas in the enterprise, primarily normalized data, and “spokes” as data marts and cubes populated from the hub but maintaining their own unique schemas.

This Microsoft Federated Data Warehouse approach leaves up to architects to decide whether to keep star schemas on the hub or spokes.

This approach combines the flexibility of data marts with benefits of Enterprise Data Warehouse’s ‘single version of truth’. It allows the enterprise to set and enforce common standards (on the hub) while answering questions that cut across business units (through spokes).

The approach is supported by fast data movements through dedicated high-speed network interconnecting all hub and spoke databases. The data can also be moved directly from node to node within the grid.

Fast Track Reference Architectures

Fast Track reference configurations target solutions on a single server estimated at up to 32TB of data and allow to scale SQL Server to 1 petabyte.

Built around the Dell and HP platforms, these Fast Track reference configurations balance all the system components—server and CPU configurations, storage designs, I/O channel configurations—optimized for BI workloads.

“Fast” definition is evidently referred to an accelerated choice of right hardware and reduced setup, maintenance, and support. It also means accelerating end user query performance, by avoiding delays or timeouts.

Tuned for Data Warehouse Query Workloads

1. As Data Warehouse activity characterized by large bulk operations with mostly disk reads, SQL Server Fast Track Data Warehouse focuses on providing the best performance for these sequential I/O.
2. The Fast Track configurations are designed to balance all tiers of hardware and software components: CPU, I/O, Storage system, and optimize all for data warehouse loads, to achieve maximum throughput capacity.

Related posts:

1. Microsoft Position on BI Market