Microsoft to develop Hadoop distributions for Windows Server and Azure
Summary: Microsoft is stepping up its support of Hadoop with new Windows Azure and Windows Server distributions in order to better support users' big data and unstructured data needs.
Context & Ripple Effects
Microsoft's move to build its own Hadoop distributions lands just days after independent year-long cloud speed tests ranked Windows Azure ahead of Amazon EC2, Google App Engine and Rackspace — a rare third-party win that gave the company credibility to push Azure beyond commodity hosting into data workloads. The distributions extend that push in two directions at once: a cloud service on Azure and an on-premises option for Windows Server shops.
The strategic logic is defensive as much as offensive: Hadoop has become the default engine for unstructured data, and until now running it meant leaving the Microsoft stack. By shipping native distributions, Microsoft keeps its SQL Server and BI customer base inside its own platform rather than ceding big data to Linux-based deployments.
First-order effects
- Enterprises standardized on Windows Server gain a supported path to Hadoop without standing up separate Linux clusters, and Azure gains a first-party big-data service to sell alongside its infrastructure.
- The existing Hadoop distribution vendors now compete against a Microsoft-branded alternative aimed squarely at the large Windows-installed base.
Second-order effects
- Rival cloud providers face pressure to match a bundled Hadoop-plus-analytics offering on their own platforms, turning big-data frameworks from products customers assemble into features platforms bundle.
- Microsoft's BI and SQL Server tools become natural integration points for the new distributions, raising switching costs for customers who adopt the stack end-to-end.
Third-order effects
- If the pattern holds, open-source data engines get absorbed into hyperscale platforms as differentiated services, with the cloud providers — not the Apache projects — controlling packaging, pricing and enterprise reach.
- That consolidation raises the stakes of the platform wars measured by the recent cloud performance benchmarks: whoever wins workloads wins the data tooling attached to them.
The trend: Cloud platforms are absorbing open-source big-data frameworks like Hadoop as bundled, first-party services, shifting control of enterprise analytics from distributors to hyperscalers.