On this first anniversary of the public release of the Azure cloud platform, Microsoft announced that it has brought 31,000 customers on board, marking a 55 percent increase from the 20,000 they claimed in July.
To mark the occasion, below are some select use-case driven headlines from HPC in the Cloud that have appeared throughout Azure’s short lifespan, beginning with this announcement introducing Azure from October 27, 2008.
- Microsoft Takes Supercomputing to the Cloud – May 18, 2010
- Siemens PLM Software to Develop on Microsoft’s Cloud – July 16, 2010
- Microsoft, Japan’s National Institute of Informatics Partner on Cloud Research Project – October 1, 2010
- Microsoft Research Makes Microsoft Biology Foundation and MODISAzure-Based Environmental Service Available to Scientists and Researchers – October 10, 2010
- Cloud-Driven Tools from Microsoft Research Target Earth, Life Sciences – October 19, 2010
- Microsoft Extends Partnerships Across Europe to Boost Scientific Research Through Client Plus Cloud Technology – October 27, 2010
- Esri Technology Brings Geospatial Analysis to Windows Azure – January 25, 2010
Over the past year, we have also featured a number of interviews with key people in Microsoft’s Technical Computing Group about the role of Azure and cloud for HPC, including Vince Mendillo and Bill Hilf.
Despite some of the use cases presented above that have ties in research or HPC, for now, a large majority of the use cases for Microsoft’s cloud offering are far more in the range of consumer-driven applications.
Among such examples are two major company use cases Microsoft released today; T-Mobile and Xerox, which both use the Azure cloud to allow family members to share updates and print from the cloud respectively. Not exactly HPC-oriented, but big names for the tech giant to gather under its cloud computing umbrella.
From these quarters, the real question is whether or not there will be a wealth of use cases emerging over the next phase of Azure growth that will have definitive value for HPC users and their plethora of applications.
Of course, one of the more interesting examples of a large-scale migration will involve watching as a mega-corporation, in this case Microsoft itself, migrates to the cloud. According to Redmond officials, they are planning a mass migration of internal systems to Azure, although there have been few details about the nature of this move—and to what degree its mission-critical systems will be reliant on it.
Speaking of mission-critical applications and their integration into the cloud, what do Microsoft’s adoption patterns look like on this front?
Recently, IT writer Tim Anderson reported on his Azure briefing with marketing manager Prashant Ketkar and asked about the number of and types of applications being deployed. While Ketkar was mum about the number (outside of saying it was “growing rapidly”) this led to some speculation about what types of applications were moving over to the cloud—and if these went beyond general consumer-driven apps.
In Kethar’s words, “No enterprise is talking about taking a tier one mission critical application and moving it to the cloud…what we see is a lot of marketing campaigns, we see a lot of spiky workloads moving to the cloud. As the market starts to get more comfortable, we will see the adoption patterns change.”
Redmond is doing all it can to differentiate the cloud services it provides from those offered by Amazon Web Services, which just announced its Elastic Beanstalk Service—a clear act of one-uppance from AWS. Furthermore, it is still trying to lead the race with IBM and Google, both of whom are seeking to bring on as many developers and enterprises as possible in this still-early stage of the cloud provider war. For now, it is finding its edge in its support of the system—although as AWS and others roll out more features for developers and users alike over the next year, it will take more than this to reel in converts.
On that support front was a news item from last week that might have resonated well for those looking for a reason to choose Azure over other options on the basis of fears of abandonment. The company produced an exhaustive guarantee against “disruptive changes” on the platform, noting that it would offer a one-year promise to notify users 12 months in advance of any dramatic changes in services that would interfere with user interactions.
As Microsoft’s David Carrington stated, for this policy, “Disruptive change broadly refers to changes that require significant action whether in the form of administrator intervention, substantial changes to the user experience, data migration or required updates to client software.” In other words, if the company suddenly decides that the Azure platform is not performing as desired or critical functions are too problematic to support, there is a written promise (which actually goes a long way in the world of cloud services since such an agreement is difficult to come by at this point) that ample notice will be provided.
The next year might be a brighter one for Azure adoption among the computationally elite, at least that is what the minds behind the Technical Computing Group hope. A new effort, which was also launched at the beginning of this new year called Technical Computing Labs is targeting developers with offerings of early versions of code, the ability to become part of the developer/development feedback process and a range of new tools in Visual Studio.
The company stated that they hope to integrate cloud computing in the process of “bringing multiple technologies and services to bear, including parallel development tools, distributed computing environments with Windows HPC Server, cloud computing on Azure and a broad ecosystem of partner applications.”
This effort also includes the launch of a new project, called Sho, which grants those with technical computing workloads and interactive environment that allows users to connect scripts with compiled code (in .NET) to allow for quick prototyping. This effort from Technical Computing Labs also includes the Task Parallel Library, which provides the basic components and algorithms for parallel computation and asynchrony.
Still, while developers are being permitted access to some attractive tools, what is needed over the coming year are a large number of HPC-relevant case studies proving that Azure has the potential to stand up to rival giants in the industry. While there have been a few notable successes, particularly in the genomics realm, finding case studies for a user population that historically (and dramatically) favors Linux might not be an easy task.
Just as we will report over the coming year on other cloud providers who are making inroads to lure in the elusive HPC crowd we will continue to watch Microsoft as it further refines it offering and produces more use cases to prove its value for a market that is can be rather skeptical to say the least.