The Xen team celebrated the final touch to over four years of work last week, when Linus Torvalds pulled xen-blkback into the Linux mainline tree. This tiny move has great consequences, because it means that the next release of Linux – Linux 3.0 as we found out yesterday – will have everything necessary to run as both as a management domain and as a Xen guest.
It's been a long journey from USENIX 2006, when IBM, VMware, Red Hat and Citrix first discussed the paravirt-ops interface. Since then, a tiny team of engineers has worked tirelessly with the Linux community, bringing new features every few weeks into the kernel, enabling first guest support, and then feature after feature towards complete control domain support. This work is now complete, and what a way to celebrate Linux 3.0!
I met Jeremy Fitzhardinge last week, architect of this epic slog, and instead of the haggard shell that I expected, I found him as bouncy and cheerful as always. I asked him how he'd survived, and he explained the debt that we all owe to Konrad Rzeszutek Wilk at Oracle, for shouldering the burden of the upstreaming effort recently. So thank you and congratulations Konrad, and Jeremy, and while I'm at it, to Ian Campbell, Stefano Stabellini, and the other people who leant their time and strength to the cause.
I was going to use this blog post to describe the history behind this in a bit more detail, but Wim Coekaerts beat me to it. Go read his post if you want to hear about the past, and I will instead talk a bit about the future.
This upstreaming is so important because it means that there is no longer any need for distros to carry different kernels for their Xen support. This is great, because it greatly reduces the burden of Xen support. One single kernel means no more Xen patches to maintain, no more forward porting, and no need for separate kernel packages. For Debian in particular – who's maintainers are all volunteers of course – this should make supporting Xen a lot easier, and will help the XCP team in their drive to make "apt-get install xapi" a reality on Debian and Ubuntu.
And once you've done all of that, you can have OpenStack on Xen on Debian and Ubuntu. And that's what makes all of this hard work worth celebrating.
By: Ewan Mellor
It's been a long journey from USENIX 2006, when IBM, VMware, Red Hat and Citrix first discussed the paravirt-ops interface. Since then, a tiny team of engineers has worked tirelessly with the Linux community, bringing new features every few weeks into the kernel, enabling first guest support, and then feature after feature towards complete control domain support. This work is now complete, and what a way to celebrate Linux 3.0!
I met Jeremy Fitzhardinge last week, architect of this epic slog, and instead of the haggard shell that I expected, I found him as bouncy and cheerful as always. I asked him how he'd survived, and he explained the debt that we all owe to Konrad Rzeszutek Wilk at Oracle, for shouldering the burden of the upstreaming effort recently. So thank you and congratulations Konrad, and Jeremy, and while I'm at it, to Ian Campbell, Stefano Stabellini, and the other people who leant their time and strength to the cause.
I was going to use this blog post to describe the history behind this in a bit more detail, but Wim Coekaerts beat me to it. Go read his post if you want to hear about the past, and I will instead talk a bit about the future.
This upstreaming is so important because it means that there is no longer any need for distros to carry different kernels for their Xen support. This is great, because it greatly reduces the burden of Xen support. One single kernel means no more Xen patches to maintain, no more forward porting, and no need for separate kernel packages. For Debian in particular – who's maintainers are all volunteers of course – this should make supporting Xen a lot easier, and will help the XCP team in their drive to make "apt-get install xapi" a reality on Debian and Ubuntu.
And once you've done all of that, you can have OpenStack on Xen on Debian and Ubuntu. And that's what makes all of this hard work worth celebrating.
By: Ewan Mellor
