While it is nice to see Microsoft playing nicely (for once) with the Linux and GPL space with their device driver for inclusion into the Linux kernel 2.6.30.1, there are some important things to keep in mind.
The code base (and I have not looked at it yet), is meant to be a device driver that will be loaded by a Linux kernel when that kernel is being run on a Microsoft virtualization platform.
Given Microsoft’s embrace, extend and extinguish track record, we have to keep constant vigilance in this set of code. I would have applauded them if the code they provided was meant to make their OS run better on a Linux KVM hypervisor. That would have meant that they really wanted to play nice. The way they have done this installment of code is for the other way around. Thanks, but no thanks. I don’t think I can find any use case where I will ever recommend any enteprise to run a Microsoft vitualization platform for any reason.
Any credible CIO and/or enterprise architect who chooses not to benefit from the high quality, highly scable, and built-in virtualization in the Linux kernel with KVM is shortchaging their organization – shortchanging in terms of having to pay more for virtualization and not being able to deploy and benefit from virtualization across the organization. Running a product like Red Hat Enterprise Virtualization with full subscription-based support is a compelling option because of the open source nature of the product as well as the stability and exnovation that is happening. Whoever controls the hypervisor, controls the destiny of best the VMs that run on it progress and perform. The IT industry cannot let a proprietary hypervisor control that and RHEV, IMHO, is the best choice to ensure that.
My sense is that even if the code is shipped in stable branches of the Linux kernel, over time, the code will go stale, because of pressures with the MS environment to focus only on their proprietary product. They will make their virtualization look and smell “better” by making the Linux VMs running on their hypervisor less than speedy and secure. Thanks, but no thanks.
So, even with MS’ GPLv2 Linux device driver code, be wary – very wary!
I think you’re correct here, this way Microsoft can claim they are a happy OSS contributor when the only value this code brings is in displacing more Linux. Further, we now have the burden of having to download all of that more often 🙂