Stefano Stabellini, a senior software engineer at Citrix Systems, has announced a proof of concept port of the open source Xen Hypervisor for the ARM Cortex A15 processor. The project was started in early November and has already developed to the point where it is capable of booting a Linux 3.0 based virtual machine up to a shell prompt. The Xen port has progressed so rapidly due to a decision to take advantage of the virtualization features that were introduced with the ARMv7 architecture making it small and comparatively easy to develop. However, because of this it won’t be able to run on anything older than a Cortex-A15 processor.
This isn’t the only Xen port for ARM underway; Samsung is also working on a Xen port that is not reliant on the ARM virtualization extensions, allowing it to work with the older ARMv5 and ARMv6 chips that do not have the virtualization extensions.
Although Stefano works for Citrix, the project is based on the open source Xen implementation, not Citrix’s own XenServer or XenClient code, and is being developed as part of Citrix’s commitment to support the open source platform. This does not rule out the possibility of Citrix developing a commercial Xen implementation for ARM at some point in the future. Citrix has not made any public statements about the possibility of a commercial release, however given the upsurge of interest in ARM-based servers it must be on the cards. Until recently this was the domain of start-ups like Applied Micro, but in recent months mainstream server manufacturers have made their first steps in to the market. HP has already launched the “Redstone” experimental server line using Calxeda’s new quad-core EnergyCore ARM chips, Dell has also been experimenting with ARM servers too.
One other point of note is the fact that the recent negative press concerning Carrier IQ will not help VMware’s goal seeking widespread commercial adoption of its own Type II ARM Hypervisor MVP (now officially called VMware Horizon Mobile). Despite much of the initial reporting on Carrier IQ being wildly inaccurate and misdirected, targeting the software’s developer as being at fault when responsibility should have been apportioned between phone manufacturers and cellular network owners. The net effect was to create the perception that Carrier IQ (and products like it) functioned as a root-kit, hidden from the users eyes, difficult or impossible to remove, sitting under the operating system and logging everything that took place. Once the presence of a root-kit is presumed, the benefits of a type II mobile hypervisor will diminish in many people’s eyes if a type I hypervisor is available as an alternative. Given the limited progress that VMware has made in getting handset manufacturers to adopt MVP – so far only LG has signed on as a possible adopter, it may now struggle to get other manufacturers on board.