KVM on Illumos (OpenSolaris fork)
dtrace.org
dtrace.org
I wasn't referring to the KVM port as a solution to an exciting new problem (it was, but a while ago, thus making the problem stale). But it was a much needed solution. I was referring to innovations that originated in the Illumos code base such as ZFS, DTrace, zones, doors, crossbow, ctf, least-privilege, etc. All of these are either unavailable on most systems, have inferior clones (i.e. systemtap, btrfs), or are partially available (the Mac DTrace port doesn't have a complete, systemic scope, due to Apple's classist tendencies, and most of the kernel providers like fsinfo are missing, for example).
Either way, I still maintain that Illumos is a nexus of systems innovation compared to other kernels, based on how much innovative technology originated in the Illumos kernel, before being ported to or reimplemented on other kernels (the Slab Allocator is an example of a technology that originated in this code-base and has been reimplemented on almost all other kernels; ZFS and DTrace are examples of technologies that have been ported over to some other kernels, albeit with less functionality/stability).
Also, the release of Illumos-KVM (among other improvements in the last year) only goes to show that Illumos is still innovating in various ways, while other kernels are either still catching up (I believe that FreeBSD didn't support DTrace'ing userland apps until their 2009 release) or losing differentiators (linux just lost KVM). Seeing as how the Illumos project has quite a bit of engineering talent and corporate investment behind it, I doubt that the rate of innovation will drop below that of other kernels... in fact it might just increase, now that potential contributors know that Oracle/Sun doesn't have a monopoly
"... a monopoly on the engineering talent."