Azure Stack is essentially a slice of the Microsoft public cloud that an enterprise can run in private, either inside its own data center or at the customer premises.
Originally, Azure Stack was intended to be software for any Windows machine. Enterprises would be able to offer their developers privately controlled infrastructure-as-a-service (IaaS) or platform-as-a-service (PaaS) based on Azure’s software, making it easy for applications to shuttle between that private cloud and Azure proper.
Microsoft found, though, that customers were more interested in a pre-assembled product, so Azure Stack transformed into a turnkey server offering, with Dell, Hewlett Packard Enterprise (HPE), and Lenovo signed on as the initial providers.
That was last summer. Azure Stack is still in preview mode, with general availability slated for the middle of this year, as Microsoft confirmed in October.
The compatibility between private cloud and Azure is important because it would avoid problems when applications get moved between the two environments. The tradeoff is that it represents a form of customer lock-in, giving the enterprise more incentive to keep developing services for Azure rather than turn to other public clouds.
This trend isn’t unique to Microsoft. Amazon Web Services (AWS), in particular, has a slew of services that would encourage users to not wander off to other clouds. Portability could also be a concern as serverless functions become more popular.
Cisco, meanwhile, has been broadening the scope of its UCS servers. The UCS M-Series is the basis for Cisco’s take on hyperconverged infrastructure. And in November, the company announced UCS-S, a modular, disaggregated design similar in philosophy to Intel’s Rack Scale Architecture. (UPDATE: The UCS M-Series was discontinued in 2016; Cisco is now pinning its hyperconverged plans on its HyperFlex products.)