There is the IaaS/PaaS/STaaS cloud, now dominated by Amazon Web Services both by monetary momentum and a long list of competitors and AWS-compatible ecosystems. Varying organizations also offer a long and interesting list of competitive services, a la carte or in various packages and formats, for pricing that varies with El Niño cycles.
And then there is Microsoft's Azure.
Azure has been emblematic of an enormous change for Microsoft, including the departure of Steve Ballmer, and the rise of forced-heterogeneity at Microsoft. Headlines of Linux compatibility and releasing their own software assets as open source would have been an anathema just five years ago. Chairs would have been thrown; secret PR campaigns would have been foisted, proxy legal battles would have been launched—a ship of lawyers to castigate and revile the commie heathen FOSS Folk.
That didn't work.
This is a different Microsoft, but no less charged with shareholder return, and using their assets to dominate markets where they reap harvests. Microsoft must play in the cloud, having lost smartphones and having paid huge prices to become somewhat tablet-ish.
The Azure Stack is something that Microsoft architects have dreamt about for several years, if only "announced" roughly a year ago. Azure took a wicked-long time to arrive not because of sloth. Rather, Microsoft had five very serious problems working against it.
Problem one was a highly constrained licensing model, along with immense relationships between server licenses, cores in the server, CALs, and various constraining factors that only an Oracle salesperson could love or appreciate for its majestic revenue model and lack of understandability. The model was built for the 1990s, when the cloud was once called "service bureau" and wasn't much competition in the burgeoning, pre-VM era.
Problem two was breaking the psychological chain of: Not Invented Here, e.g. embracing other products that might actually exist and products that might be, competitively speaking, diametrically opposed to Microsoft's client domination and revenue models. Microsoft learned a little from Apple's direct-to-consumer marketing and the affinity and (yes, maybe) love that it can purchase for client stickiness.
Problem three was an architectural model that allied and promoted on-prem, data center-focused infrastructure models. The Principality model of commerce that Microsoft had adopted didn't match the Hanseatic League federation model that the cloud brought forth. Microsoft feared losing clientele if they moved from backward-compatible API sets that were designed to please accountants who wanted long depreciation on hardware and software assets. Times change.
Problem four was losing developer love. After Microsoft started attempting to market directly to one of their cores – third-party developer organizations – developers started finding themselves in the dust of agile development, the DevOps mentality, open source principles, and those that might embrace not-invented-here so as to just get their increasingly demanding jobs done.
And the final problem was perhaps the most onerous, the fact that the world no longer revolved around Microsoft, no longer were there lines of people waiting outside of a store, breathlessly holding fistfuls of cash for a new release. Microsoft really had no secrets, and no rumor mill and market panache that had been happily appropriated by Apple (which faces similar issues of ho and hum).
So Azure in your data center matters, because it greases the wheels for you to use Azure Cloud, which has been constructed to new standards at Microsoft — a different, more scarred Microsoft, having learned the lessons of disastrous Windows 8 product launches after the wide embrace of Windows 7.
Microsoft has spent a good amount of time to try to perfect Azure and stand it up to its perceived competition. There are good infrastructure designs. Plentiful glue for hybrid networks without making hefty investments in parallel infrastructure engineering on the part of its clientele. It's almost embarrassingly (for Microsoft) open. It's ready for work.
But are Microsoft's clientele ready to dive in to the available models—or were the early and even not-so-early adopters from free-form cloud ready to taste Azure? Time will tell, once the final bits (there always seems to be final bits undone) arrive on the scene and make it to 1.1.