Drastic measures are needed if Microsoft is to make an impact on Amazon’s cloud control, says Peter Judge
Microsoft’s cloud vice president Satya Nadella has hubris. “I believe we will be at the core of, and in fact lead, the enterprise cloud era,” he proclaimed before yesterday’s big cloud announcement of upgrades to Azure, the Microsoft cloud which has negligible market share and is prone to embarrassing high profile failures.
To be fair, Microsoft has some good cards to play. Its server products are well established, and it’s a major enterprise software vendor, so it ought to be possible to turn some of that business into a greater cloud presence. Microsoft also traditionally does well in government contracts, so its move to launch a Government Cloud in the US is a very canny one.
Windows in the cloud?
But Microsoft’s Azure has so far done so little business that it doesn’t register outside the “other” group in 451 Research’s market share figures. It’s far behind Rackspace and lagging comparative unknowns like Joyent and CSC in a market which is utterly dominated by Amazon.
Microsoft has promised to match Amazon’s prices for infrastructure as a service (IaaS), but Azure was always intended as a platform as a service (PaaS) product, offering a cloud version of Windows. It’s now getting more serious about IaaS, but, as analyst William Fellows of 451 Research says: “If 70 percent of Azure footprint is now IaaS as Microsoft said, it must have had a very small PaaS uptake.”
It’s also had a couple of very embarrassing failures. In February this year, users got the Azure screen of death, when Microsoft’s cloud failed to handle the expiry of an SSL certificate. More embarrassingly, on 29 February 2012, Azure stopped working because Microsoft failed to take account of the leap year.
The challenger needs something different
Microsoft is used to having a stranglehold on its major markets, but in cloud, as in mobile devices, it is an upstart. As such, it needs to offer something different. And that something could be openness, analyst Clive Longbottom of QuoCirca suggests.
Amazon’s web offerings are effectively de facto standards, he says: “Everyone has created AWS/EC3 connectors, so it is relatively standard at the edges. However, you can’t take an AWS image and run it elsewhere – it would need to be recompiled (or possibly even recoded).”
This lack of openness could be a weakness of Amazon, if rival cloud providers opened up. There are credible open source cloud offerings from the Apache CloudStack project and from OpenStack, which enjoys heavy backing from Red Hat.
At the moment, Microsoft’s Azure is still just as proprietary as Amazon – but with a tiny user base it will never be seen as a de facto standard. For those who want something more open than Amazon, Cloudstack is seen as the best alternative, says Longbottom – but that’s a role which Microsoft could grab if it opens open up its own code.
“Microsoft should open up Azure and make it available as a platform to other service providers,” says Longbottom. “Keeping Azure to itself is harming Microsoft.”
An open cloud implementation is not a whimsical choice either. Services based on OpenStack can be easily moved between public and private clouds, and the user can choose where to run a given workload in the “hybrid” model which Microsoft, like every other provider, says it espouses.
Making Azure open would be a brave step. But trailing as it is in cloud, Microsoft has to do something pretty dramatic to make a real impression.
Do you know Microsoft? Try our quiz!