Hyper-V shines spotlight on CentOS, ready soon for 6.0 release
What's happening this week in enteprise Linux? Quite a bit.
Dueling announcements from the Open Source Business Conference earlier this week involved Red Hat in virtualization space, with one announcement setting up Red Hat's defense. The other demonstrating exactly why Red Hat needs a defense as a spotlight shined on a significant Red Hat competitor... a competitor that may soon be making a big release announcement of their own.
First, the defensive strategy.
Today's announcement of a new industry consortium called the Open Virtualization Alliance (OVA), was total reaction to the power of VMware in this space. Rather than watch VMware bully its way to the top, seven (count 'em) seven companies with big stakes in keeping the virtualization sector's playing field level BMC, Eucalyptus Systems, Hewlett-Packard, IBM, Intel, Novell, and Red Hat gave birth to, er, OVA.
The new consortium will focus on the continued development and deployment of the KVM virtualization platform, which is good news for Red Hat, which uses KVM for its virtual toolset. OVA members hopes the new consortium will attract a lot of smaller vendors who don't want to get run over by VMware or the Cisco-based Xen technology.
The news came as a bit of surprise, at least because of the presence of HP, IBM, and Intel. IBM has a significant investment in supporting Xen, though according to The Register, "all new development will be on KVM."
Getting the upper hand in the virtualization space is very, very important, since the better virtualization platform will have the edge on revenue from the massive data centers that will be the basis of the cloud. This is why we will see a lot of maneuvering in this sector.
Red Hat, which bet big on KVM over Xen, wants to make sure it can ride that bet all the way to the payoff. I'm sure they did a lot of deal-making to get these particular players involved in OVA, because if they are deeply integrated with the virtualization layer, they won't find themselves vulnerable if a virtualization vendor decides to get squirrelly.
And hey, Sherman, Red Hat only needs to set the Wayback Machine for a mere 24 hours in the past to see an example of such squirreliness.
At the same conference on Monday was the other piece of virtualization news: Microsoft announced that CentOS would be an optimized OS on its Hyper-V virtualization platform--specifically Windows Server R2 Hyper-V.
This, then, would be the reason why Red Hat feels it needs a strong KVM environment.
According to the announcement, Hyper-V users can not only run CentOS smoothly, but Microsoft will actually help said customers with CentOS configuration.
The uninitiated to the nuances of Linux politics might be tempted to think "gee, how nice of Microsoft to support Linux. Isn't that nice?"
As a public safety announcement, I urge people who actually know what's going on not to slap such people upside the head.
Instead, politely inform them that by supporting CentOS on Hyper-V, Microsoft is trying to validate CentOS and hurt the commercial distro upon which CentOS is based: Red Hat Enterprise Linux (RHEL). A big detriment to Red Hat in the commercial market is not just competing Windows and UNIX systems, but also the presence of non-commercial clone distros like CentOS and Scientific Linux, which offer very much the exact functionality of RHEL without the expensive support costs.
The cloning situation was bad enough that the Red Hat development team took the unprecedented step this spring to obfuscating the source code of its kernel by releasing that code in one big tarball file, with all of the patches applied already. Functionally, this made no difference for RHEL users, since the kernel is optimized for, well, RHEL. But for downstream projects like CentOS, Scientific Linux, and Oracle Linux, it is now much harder to discern the exact changes Red Hat has made to its kernel and therefore make changes optimized for their downstream distribution.
At the time, Red Hat was adamant that this move was to combat its commercial competitor, Oracle.
"When we released RHEL 6 approximately four months ago, we changed the release of the kernel package to have all our patches pre-applied. Why did we make this change? To speak bluntly, the competitive landscape has changed. Our competitors in the Enterprise Linux market have changed their commercial approach from building and competing on their own customized Linux distributions, to one where they directly approach our customers offering to support RHEL," Stevens wrote in March.
"Frankly, our response is to compete. Essential knowledge that our customers have relied on to support their RHEL environments will increasingly only be available under subscription. The itemization of kernel patches that correlate with articles in our knowledge base is no longer available to our competitors, but rather only to our customers who have recognized the value of RHEL and have thus indirectly funded Red Hat's contributions to open source that will advance their business now and in the future," Stevens continued.
Of course, I'm sure Red Hat wasn't shedding a tear over the fact this move would harm it's non-commercial clones, too.
This puts Monday's Hyper-V announcement in a broader context, because now Microsoft is sending business to CentOS... and therefore away from Red Hat. Sure, they have to grit their teeth and support Linux, but if they can weaken the source of CentOS enough, they imagine they can chop off the head and kill the entire RHEL ecosystem.
Yeah, good luck with that.
While CentOS is a popular non-commercial distro in the enterprise, it's not been entirely clear how CentOS itself is doing. The team just released CentOS 5.6 on April 8, 86 days after the release of of RHEL 5.6 on January 13. That's not too bad, but the RHEL 5 kernel code is not under the new kernel obfuscation plan; that's a RHEL 6 thing.
So where are we with CentOS 6? As of today, May 18, its been 190 days and counting since the November 10, 2010 release of RHEL 6. Oracle Linux 6, by comparison, was released in 94 days (Feb. 11) and non-commercial distro Scientific Linux saw its 6.0 release 114 days (March 3) after the RHEL 6 release.
I put the question to CentOS lead developer Johnny Hughes, who responded with some details on what's happening:
"CentOS 6 is getting very close. The main issues that we are having with CentOS 6 is caused by two things. The first is that we can not build on our current build system since there is no interoperability of RPMs built on EL5 and EL6 (see this bug)
"Our current build systems that have CentOS-5 on them can not be used because of this, so we needed to design a whole new build system for CentOS-6. Since we wanted to build on CentOS and not the upstream EL6 distribution, we needed to literally build what we needed to create the build system first.
"The second issue we are having is that not all the files that Red Hat used in the build process are released. They have intermediate files on their 'Staged' build system that were newer than their Beta release of EL6 or Fedora 12, but older than those released in EL6. These were used to build the released RPMS, but they are not actually in EL6. We have to find (or build) versions of these missing files to get the proper builds.
"We have a CentOS-6 tree with about 95% of the packages to our QA testers now, I expect we should have a release in a couple of weeks."
Which will be welcome news indeed to CentOS fans. Hughes disputed any assumptions that Red Hat's kernel obfuscation plan has adversely affected CentOS.
"As far as the kernel source code, no that has not impacted our ability to build at all. It would impact our ability to pull out a specific patch for testing if we were troubleshooting a problem... like Troy Dawson of Scientific Linux did with this GLIBC issue," Hughes wrote. "But as far as just rebuilding the kernel goes, there is no impact."
I also asked Hughes how the CentOS team reacted to the Hyper-V announcement from Microsoft. He replied:
"The CentOS Project is happy to be included as a supported OS by Microsoft in their Hyper-V system, but I do not agree with the statement by The Register that CentOS is now a 'First Class Citizen.' We have been a 'First Class Citizen' in the Linux world for quite a while. CentOS is currently the most used Linux distribution on the top 1 million Web servers on the Internet...
"Our webserver usage is more than Red Hat Enterprise Linux and Fedora combined as measured by W3Techs.
"I am sure the reason that Microsoft picked CentOS to support is that they were looking for an Enterprise level product that was available with no added cost for their customers. CentOS meets that requirement. Here is a blog where they explain why they picked CentOS (it was
because of our market share and at the request of their customers, not to 'get back at' anyone else).
"With respect to Microsoft, as long as they abide by the OSI licensing requirements for any software they distribute, the CentOS Project is fine with them distributing and supporting CentOS for their customers."
So Hughes believes the advantages of his distro and its free-from-cost status are the big reasons why Microsoft chose to support CentOS on Hyper-V. While you can't take that away from him, I can also see Microsoft playing both ends against Red hat in the middle: get a free enterprise distro, and put the screws to Red Hat at the some time.
All of this Hyper-V/CentOS drama underlines my original point: Red Hat must keep a strong stake in the virtualization platform, if only to avoid dramas like this, where a vendor that doesn't like Red Hat (and there are so many) opts to support Red hat's competitors.
Better for Red Hat, then, to keep pushing KVM as far as they can so they will always have a safe haven in virtual space.