On October 22nd, Microsoft announced that it has partnered with Cloud.com to provide integration and support of Windows Server 2008 R2 Hyper-V to the OpenStack project. The announcement caused a great deal of interest here at the Virtualization Practice, as it signals an unexpected willingness on Microsoft’s part to pursue interoperability at the IaaS layer, allowing users to break out of the Hyper-v stack, whilst still retaining Hyper-v at the bottom. The fact this announcement came from Microsoft (not Cloud.com, Rackspace or OpenStack) seems to signal the seriousness of the intent.
In practical terms this means that Cloud.com puts a Hypervisor Abstraction Layer into the bottom of the OpenStack compute platform (Nova), and binds Hyper-V into that, to allow images to be deployed to and controlled on Hyper-V from OpenStack, using tooling that speaks one or other of the two OpenStack APIs (Native or Amazon EC2). Technically it is not a major step because although the initial version of Nova targeted libvirt and thereby Xen, KVM and Qemu, Citrix had already succceeded in providing a hypervisor abstraction layer in OpenStack for XenServer.
The initial cut of the Hyper-v code missed the release date (25th October) for the recent “Austin Release” of OpenStack, but is present in the nightly build if you want to try it, and is due to be formally released in the Bexley Release in January. It’s not Microsoft-supported product, but it has been built with their help. The Hypervisor Abstraction layer at this stage is a “least common denominator” internal API, but over time we may expect API consolidation or API pass-through to allow use of the Microsoft System Center tools developed for the Hyper-V cloud against the Hyper-V instances, in a Microsoft/OpenStack/Microsoft sandwich, where Microsoft provides the management tooling and the hypervisor but not the base cloud infrastructure.
We do not really know if the approach is strategic for Microsoft, but we have some insight into the genesis of the project. The initiative came from Microsoft customers who were looking for an Open IAAS cloud layer, but who had already standardized on a Hyper-V virtualization platform. Likely the driver was some kind of public cloud compatibility requirement or a cloud-bursting requirement, which can be delivered by OpenStack through API compatibility with either Amazon or Rackspace (#1 and #2 in the marketplace). There end up a three-way conversation involving Rackspace, Microsoft and Cloud.com, and Cloud.com agreed to build the Hyper-V layer for OpenStack.
Note, however, that the direct benefit to Cloud.com is very limited because it does not currently consume the OpenStack compute layer (Nova) in product. As we indicated in an earlier post, Cloud.com has its own version of the Compute infrastructure called CloudStack, which it is actively selling into the marketplace and it considers to be a more mature software offering at this stage. It plans to use some of the codebase developed for OpenStack in its own product, and over time we may see it focus on tooling and value-add which is portable between its CloudStack base and the OpenStack base.
In the meanwhile 260 people from a huge range of companies (including Microsoft) are currently in San Antonio mapping out the future of OpenStack technology, for the next two quarterly releases. It is not yet in a state where your average IT shop should be downloading and installing it, but it starts to look like it can be the Interoperability Layer in the IAAS stack, with vendors adding value above and below the abstraction it provides. Citrix is already playing the game, offering XenServer as the underlying hypervisor, and committing to using OpenStack in the Citrix OpenCloud, but the key question is how VMware responds, and in the hosting provider marketplace are also curious about whether Parallels will view this as an interesting opportunity.
We are continuing to track this project, including the governance issues we raised in a previous post and will provide you with updates over the coming weeks.