Oracle made a big noise in the Linux community yesterday by announcing its own spin on the Linux kernel on top if its so-called Unbreakable Linux. Oracle presented the announcement as offering a "modern" Linux kernel on top of its own clone of Red Hat. Underneath the hype, what's Oracle really offering, and what does it mean for Linux?
For years, Oracle has ridden Red Hat's coattails and tried to present it as a good thing to its customers. Oracle rebrands Red Hat Enterprise Linux (RHEL), slaps its own support package on it, and lets Red Hat do all the heavy development lifting while it tries to poach Red Hat's customers. All perfectly legal according to the licenses that RHEL is shipped under, but a bit skeezy nonetheless. Or perhaps parasitic might be a better word.
Don't get me wrong, Oracle does contribute to kernel and other open source development. In fact, Oracle was one of the top 20 employers by kernel contributions from the 2.6.33 kernel (as measured by Greg Kroah-Hartman). Specifically, Oracle was responsible for about 1.3% of the changesets in 2.6.33, just after AMD with 1.6%, and kernel heavyweights Texas Instruments and Fujitsu (1.9% each), and Nokia (3.0%). It's far, far behind Red Hat's 11.6% and even IBM's 4.8%. My former employer, Novell, has slipped a few notches with just 3.7% of kernel contributions — but still well ahead of Oracle. And, of course, that's only kernel development — it doesn't measure the contributions that companies put into the GNU tools, X.org, and other vital components of a Linux distribution.
Back to Oracle's big announcement. The big news from Oracle is that it's offering a "modern" Linux kernel that's supposed to offer better performance and support for newer hardware (like solid state disks), and is optimized for Oracle hardware and software. So, what's Oracle doing to perform this miracle? They're shipping a kernel based on the 2.6.32 mainline Linux kernel and comparing it to the 2.6.18 series kernel that RHEL 5.x supports. That's it, that's the big announcement — Oracle is simply offering a newer kernel than Red Hat and fine-tuning it for Oracle's own software.
Oracle also touts data integrity extensions and better support for power savings. Again — all true, but also Oracle attempting to take credit for nothing more than the passage of time and continuing kernel development and improvement. Most of the improvement done by companies other than Oracle.
It will be interesting to see, once Oracle's kernel is actually out in the wild, how it performs under workloads other than benchmarks specifically targeted at Oracle's workloads or hardware that was not well-supported in the 2.6.18 kernel series. It will be doubly interesting to see how Oracle's 2.6.32-based kernel performs against the kernel in the next RHEL release.
Red Hat Enterprise Linux 6 is just around the corner, and should feature a kernel based (loosely) on the 2.6.33 Linux kernel with some bits from 2.6.34 and probably 2.6.35 as well. So customers with RHEL subscriptions will get pretty much all the goodies that Oracle is touting, plus all the updates that go into the rest of the distribution.
However, many companies may not want to move their systems to RHEL 6, so this move may appeal to some customers. If you can upgrade in place to a faster Linux kernel on top of RHEL 5, it saves a lot of migration effort for companies that have standardized on RHEL 5 already. Red Hat may want to think about offering a similar upgrade.
At the same time, Oracle is just reaffirming that Red Hat is the gold standard (at least right now) for Linux on the enterprise side. While the company is trying to sway customers with its special sauce kernel, it's also promising compatibility with RHEL 5 and trying to leverage all the ISV support that Red Hat has.
Is this good for Linux? I'd say it's ultimately neutral to negative. Oracle is doing as little as possible to create real value in the Linux community, while trying to siphon business away from one of the few companies doing the heavy lifting. Oracle does do some original development — it sponsors Btrfs development, for instance — but it's doing the bare minimum to push its own agenda.
At the same time, Oracle is busy buttoning up Solaris development once again and trying to convince customers that they really want to be using Solaris for mission critical software deployments. If Linux takes a hit, Oracle can hope to capitalize on that by trying to lure businesses back to Solaris.
When all is said and done, I don't expect too many organizations to buy into this. Oracle is trying to create a top-to-bottom stack of offerings based on its hardware, operating system, middleware, and applications. This is the kind of lock-in that any sane IT manager, CIO, or CTO would avoid like the plague. While the economy has been slumping along and companies have been pinching pennies and cutting staff, Oracle has raised its prices. This was before it was also a hardware vendor.
The bottom line: Oracle's kernel is much ado about very little. They've tweaked and tuned a newer mainstream kernel, stuck it on top of another company's Linux distribution, and are touting it as a major accomplishment. The magic ingredient here is spin, not innovation.