The HPC Side of the Microsoft-Oracle Partnership
Microsoft and Oracle surprised the cloud computing space yesterday by announcing a partnership that would offer Oracle software on the Azure platform. According to Microsoft Chief Executive Steve Ballmer, the two, who have been known rivals for years, have been actually working together behind closed doors for a time.
“In the world of cloud computing, I think behind-the-scenes collaboration is not enough,” Ballmer explained.
The technical side of the announcement does not amount to anything particularly significant, especially not on the HPC side, as Oracle software will be run on the Azure platform per details of the deal. Perhaps the most compelling bit of technological news is that Microsoft is allowing Oracle’s Linux-based Java applications to run on Microsoft servers in Azure.
Considering Microsoft’s effort to maintain a flexible IaaS in their cloud offering, this allotment could make sense. “Since Azure now essentially runs virtual machines and provides Infrastructure-as-a-Service, it is effectively a cloud operating system,” according to IDC analyst Al Hilwa. “It only makes sense to see other popular technologies that need an operating system run on Azure.”
What will be more interesting to follow is how Microsoft and their cloud offerings in Azure appeal to the market base with the addition of Oracle software, and what a potential deeper relationship could mean. Specifically, a great deal of the HPC applications run in the cloud are on Amazon Web Services, especially with regard to scientific applications. The notion here is that adding Oracle software to the mix would cut into that.
Cowen & Co. analyst Peter Goldmacher is not so sure, saying that “we believe developers evaluating Azure will be much more likely to select Microsoft’s own SQL Server or other low cost data management technologies like Hadoop (also supported on Azure).” Essentially, those who would be experimenting with HPC on Azure have other options that are more logistically compatible, such as Hadoop.