Microsoft has joined a new project to accelerate the development of ARM-based servers, suggesting ARM versions of products like Windows Server and Hyper-V could be in the works.
Microsoft has participated in the development of a specification to help standardize the ARM server platform, so that software developers can create products for ARM with the knowledge they'll run on servers from a variety of manufacturers.
Other companies that signed onto the effort, announced Wednesday at the Open Compute Project Summit in San Jose, California, include Linux developers Red Hat, Suse and Canonical, server makers Hewlett-Packard and Dell, and chip vendors AMD, AppliedMicro, Cavium and Texas Instruments. Citrix, which makes the Xen hypervisor, is also on board.
Microsoft's participation is interesting because it hasn't discussed any plans to develop ARM versions of its server software, though it offers an ARM version of its client OS for tablets, Windows RT.
ARM said that the first version of the specification, called the Server Base System Architecture, has been completed and would be posted on its website Wednesday.
"We can confirm Microsoft participates in an industry forum driven by ARM around the Server Base System Architecture, as a part of our continuous work with industry partners to deliver solutions our customers want," a spokesman for Microsoft said via email.
He declined to comment further on the announcement, including whether Microsoft is developing server software for ARM systems.
Momentum is gathering around ARM server chips, which are seen as a low-power alternative to x86 suitable for various Web and data analytics workloads. Much of the software developed for those systems today is open source.
One of the strengths of the platform is that ARM, a U.K. chip design company, licenses its CPU architecture to numerous companies, creating a thriving market for ARM chips. But that also creates the potential for fragmentation, so ARM needs to ensure a degree of compatibility without preventing its licensees from innovating around its chip designs.
"One of the balancing acts you have to play is around standardization versus innovation and choice," said Lakshmi Mandyam, ARM's director of server systems and ecosystems.
The specification aims to address that by defining basic requirements for ARM SOCs (systems on chip), so that if a chip maker builds an SOC with a particular hardware accelerator, for instance, the software will be able to discover it and take advantage of it. The spec also standardizes the features that hypervisors need to target, such as the I/O components built into ARM SOCs.
"It's also important from a deployment perspective, because end users who deploy hundreds or thousands of ARM servers need to be able to apply new patch sets and software upgrades -- all that stuff has to be simple, unified and manageable," Mandyam said.
Ian Drew, ARM's chief marketing officer, introduced the spec in a talk at the Open Compute Project Summit Wednesday, saying it addresses a "tremendous problem."
The problem is that multiple software companies would be developing multiple OSes for multiple different ARM SOC implementations, he said. There would also be a lot of wasted effort, as chip makers designed their own versions of basic components such as timers and memory controllers.
"For ARM and its ecosystem, it needed to be fixed and it needed to be fixed now," he said.
The spec addresses firmware, hypervisors and operating systems, but it's a hardware spec rather than a software spec, Drew said, written by hardware and software developers together.
"We've released it today and it covers OSes from Linux through Microsoft," he said.