It seems that the original subject of my post yesterday caught the eyes of much of the virtualisation community, including Simon Crosby, formerly from Xensource, and now working for Citrix.
He’s written a typically well thought out response, which covers off a lot of points:
- HP have a multi-hypervisor management tool already which signs off on Xenserver, VMWare and Hyper-V support
- Xenserver Platinum, which is comprised of Xenserver Enterprise and Citrix Provisioning Server, can already provision VMs to not only physical hardware and Xenserver, but to other hypervisors as well
- He covered off again the ecosystem building around the Xenserver product range, specifically in HA areas – products like Marathon Everrun and Stratus Avance.
He also wrote up a good bit on the position of Xen with regards to KVM. I haven’t really looked into KVM much, due to not ready access to test hardware with VT capable chips (the test hardware I do have is tied up with testing Xenserver), but I’ve always been wary of various claims that it’s a better VM stack than Xen is. (That might just be because I’ve not spent the time looking into it, and it might be because of the general not-invented-here feeling the “linux kernel” community seems to have about Xen. Again, not something I’ve spent a lot of time on). A lot of the stuff Simon writes is high level and enthusiastic of course, but it paints a clear picture – Xen already has massive uptake in mindset, and not just with traditional linux vendors either (Sun xVM and Oracle VM having Xen based stacks as well). I guess the jury might still be out on which technology actually is technically superior, but as history demonstrates, it’s not always the technically superior technology that lasts.
Simon also claims that Xen will be in the BIOS hypervisor offering from Phoenix, which is something I haven’t heard before. It certainly makes some amount of sense for Phoenix to not rewrite an entire hypervisor stack and then stick it some place that’s inherently difficult to upgrade – your BIOS, but I’m not sure how it works out regarding Xen’s requirement for a privileged Xen-aware guest to provide hardware drivers.
Simon also makes another point that I must have heard before from him, because it’s stuck with me and I agree entirely with the premise:
The founding thesis of XenSource, and the continued strategy at Citrix, is to promote fast, free, compatible and ubiquitous hypervisor based virtualization. If the hypervisor is free, why worry about who delivers it? Let the customer pick the implementation method that they want – the real money is in the up-sell with products that make virtualization valuable for customers.
Whether you like it that companies are in this to make money or not, this approach seems a good one. Piggybacking their moneymaking on an opensource product, an action which drives development, acceptance and that horrible word “mindshare”, doesn’t have to be a bad thing.