Red Hat acquires Qumranet [LWN.net]

Red Hat announced that it has acquired Qumranet, Inc., developers of the KVM virtualization tool. "Qumranet is the inventor and key maintainer of KVM, the only virtualization technology that is fully incorporated into the Linux kernel. Red Hat views KVM as the next generation of virtualization technology -- it combines support for the latest hardware virtualization capabilities and the rapid feature development of the Linux kernel into a complete, highly functional, virtualization platform. Red Hat believes that a strong coupling between the hypervisor and the kernel is a major advantage." (thanks to Matt Domsch).

to post comments

Red Hat acquires Qumranet

Posted Sep 4, 2008 15:10 UTC (Thu) by drag (guest, #31333) [Link] (1 responses)

> Red Hat believes that a strong coupling between the hypervisor and the kernel is a major advantage.

It is. This is terrific news, I think Redhat scored big with this one.

Lets not forget their other products either. Such as 'Solid ICE' for running thin clients that connect to virtual machines.
http://www.qumranet.com/products-and-solutions

It'll be interesting to see if anything happens with that.

License for SPICE

Posted Sep 4, 2008 19:35 UTC (Thu) by dmarti (subscriber, #11625) [Link]

If previous RHT acquisitions are any guide, it's likely that they'll open source the "SPICE" implementation (which they claim is like RDP, only fast enough to handle video.)

Open Source acquisitions

Posted Sep 4, 2008 15:12 UTC (Thu) by alex (subscriber, #1355) [Link]

It certainly makes sense, KVM is rapidly winning mind-share over Xen type approaches and has a better track record with working with the kernel community.

I'm guessing the acquisition process is a lot easier with Open Source companies. The technology is in the open and there is no bar to working with them while you do your due diligence. However I wonder if it implies there will be a relatively static number of big players with deep pockets who hoover up teams of engineers around a certain technology?

Maybe also protective move

Posted Sep 4, 2008 16:16 UTC (Thu) by kragil (guest, #34373) [Link] (1 responses)

Red Hat is now a lot more undesirable for Vmware. ( If they ever were that is, but speculations seemed to suggest so. )

Maybe also protective move

Posted Sep 4, 2008 17:26 UTC (Thu) by SEJeff (guest, #51588) [Link]

For redhat to be bought by EMC (owner of VMWare) it would have to be a hostile takeover. Redhat would not take lightly to that.

Red Hat acquires Qumranet

Posted Sep 4, 2008 19:20 UTC (Thu) by hisdad (guest, #5375) [Link] (2 responses)

Its not the coupling between the kernel and the kvm that's the issue.
Its the coupling between the guest os and the hypervisor.

How much do you patch your guest to get good performance??

Red Hat acquires Qumranet

Posted Sep 5, 2008 2:40 UTC (Fri) by drag (guest, #31333) [Link] (1 responses)

Well the code you modify in the guest will be in the form of paravirtualized drivers.

So for Linux clients they are slowly integrating PV drivers as KVM matures. So there are PV drivers for block devices and network drivers and memory balloon drivers (for reclaiming memory from VMs.

http://kvm.qumranet.com/kvmwiki/Virtio

For example the fastest non-PV device for KVM is the Intel 1Gb/s ethernet adapter. When I moved to a PV driver increased the network performance of the virtual network by 300% and lowered the cpu usage in both the guest and host.

Then there is only a network driver for Windows for the same thing.

Not only they are faster

Posted Sep 10, 2008 17:05 UTC (Wed) by gwolf (subscriber, #14632) [Link]

As virtio drivers don't have to interact with real-life hardware, only forward the requests to the hardware interfaces in the host OS, virtio drivers are also much simpler - they will probably be more secure.

SolidIce and the SPICE protocol?

Posted Sep 4, 2008 19:31 UTC (Thu) by dowdle (subscriber, #659) [Link] (13 responses)

I'm sure it will take Red Hat some time to sort through the product line and code. Qumranet, while the sponsor of the development of KVM, was NOT really an Open Source company. Their business model is/was based completely on proprietary products (accept for the VDS running Linux/KVM). SolidIce is primarily a solution for virtualizing Windows Desktops... and I believe all of their management tools were Windows only (but I could be wrong about that). Their SolidIce product is definitely closed source and their SPICE protocol is patented.

This begs the question... will Red Hat continue with its commitment to Open Source and vet the code to see how much work it would be to GPL? Will they release what they can (probably all of the code unless they licensed some pieces from others) under the GPL? Will the donate the patent(s) on SPICE to the community patent pool (what was the name of that again?)?

Does this mean that Red Hat will continue developing SolidIce and sell it... or will they integrate it into oVirt? Or will Red Hat sponsored oVirt development slow down and eventually stop?

Them getting the core KVM developers is a big win... but I look forward to future announcements about the other pieces they got along with them.

If you haven't checked it out yet, see this SolidIce demo video (in Windows WMA format only - hopefully Red Hat will eventually change that):

http://us.qumranet.com/videos/Qumranet.wmv

While they only show virtualization of Microsoft Windows desktops, it supposedly would work equally well for Linux desktops. What you can use as a smart terminal device is still unclear to me.

SolidIce and the SPICE protocol?

Posted Sep 4, 2008 22:59 UTC (Thu) by rahvin (guest, #16953) [Link] (11 responses)

Redhat's history indicates they will open source all the code they own and will rewrite and open source the stuff that they don't. To deviate from this history would be a mistake and I personally don't see it happening.

SolidIce and the SPICE protocol?

Posted Sep 4, 2008 23:24 UTC (Thu) by jspaleta (subscriber, #50639) [Link] (10 responses)

The more interesting question is....
will Red Hat open source these acquired codebases faster than Canonical opens up Launchpad?

-jef

SolidIce and the SPICE protocol?

Posted Sep 5, 2008 11:55 UTC (Fri) by Lovechild (guest, #3592) [Link]

Given that Launchpad was put into default use in 2006 (Dapper LTS) that gives RH a few years to met your challenge.. or put in other words, I will eat a Red Hat style fedora whole if they fail to.

Red Hat have always played nice with us, showing a true understanding for Free Software and the way the community works, it would be a major surprise if they suddenly stopped.

SolidIce and the SPICE protocol?

Posted Sep 5, 2008 13:09 UTC (Fri) by SEJeff (guest, #51588) [Link] (8 responses)

Jef "I like to troll canonical and Ubuntu on LWN" Spatela,

Please STOP with it already. We have become quite aware that you aren't a fan of canonical or ubuntu by any means and get it. You realize that Fedora uses upstart, developed by a Canonical employee, right? You realize that Canonical developers have upstream commit access to system-config-printer because they submitted a ton of bug fixes, right? You realize that you make yourself (and fedora) look childish by acting the way you are. You might also point out you are a longtime and respected Fedora community member before trolling. That might, just perhaps, make you look better... NOT.

Please stop being childish. You can email me from my website if you'd like to continue this privately.

---
Jeff Schroeder
http://www.digitalprognosis.com

SolidIce and the SPICE protocol?

Posted Sep 5, 2008 16:35 UTC (Fri) by jspaleta (subscriber, #50639) [Link] (7 responses)

It's childish to put the concerns over Red Hat's plans for new acquired closed source software as a product line in context with what other companies are doing with regard to closed codebases which are important to their business plans? Really?

Are Canonical's Lauchpad and Landscape any less important to discuss in the context of a wider linux community compared to RedHat's newly acquired SolidICE or recently opened Spacewalk? Both Red Hat and Canonical are corporate entities which are essentially looking to make money from a services model support linux distribution. I think its perfectly reasonable to contrast how each of those companies are handling the access to the codebases behind the products and services they provide.

In fact, I think LWN should do an entire editorial series on the future of openness as applied to web services in general. This is a very important topic, especially if the "Online Desktop" concept gains traction and we as end-users are making heavier and heavier use of services in the cloud and less use of client-side code. Whose out there championing the cause of open web services and who isn't?

Maybe you aren't aware of this but Canonical != Ubuntu. The Ubuntu community doesn't control the Launchpad codebase...Canonical does. The Ubuntu community doesn't control the Landscape codebase.. Canonical does. I have concerns about Canonical as a corporate entity and the decisions being made at that level. Not it engineers, not the wider volunteer Ubuntu community.

If you want to hold upstart up as an example of how things should be done. Then great. It is a great example. Launchpad should be as open as Upstart so that non-canonical employed people can contribute to Launchpad and make it better...just as people are doing now with upstart and dbus integration in upstart 0.5. Considering that Launchpad is meant to be a tool by which Ubuntu developers can work more closely with upstream projects..you'd think it would be in everyone's best interest so that those upstream project could contribute back to Launchpad to optimize the workflow to help those upstream project developers.

I've made absolutely no claim that individuals who work for Canonical don't understand how this should work. But I'm not sure Canonical's management team gets it. Why Upstart but not Launchpad? Why isn't Launchpad open to the same community innovation that upstart is? Maybe its because Canonical's management team isn't actually convinced that the open source model is profitable and they see Launchpad as 'important'..and Upstart as ultimately unimportant. Who knows... Canonical is a privately held company so there is no transparency with regard to its business model.

If people want to question Red Hat's commitment to giving back to the open source ecosystem, then I expect then to hold up each and every other corporate entity in the community to the same scrutiny....Canonical included. Fair's fair.

-jef"Wow you mispelled my last name"spaleta

SolidIce and the SPICE protocol?

Posted Sep 8, 2008 17:26 UTC (Mon) by drag (guest, #31333) [Link] (1 responses)

> It's childish to put the concerns over Red Hat's plans for new acquired closed source software as a product line in context with what other companies are doing with regard to closed codebases which are important to their business plans? Really?

FFS. Do you have any idea how many closed source products Redhat has obtained via purchasing then turned around and open sourced them?

A few off the top of my head:

Redhat purchases Sistina. Now we have completely _Free_software_ cluster aware file system and clustering suite from Redhat.

Redhat purchases 'Netscape Enterprise Suite' products from AOL; namely Netscape Directory Services and Netscape Certificate Management System. Now we have completely _Free_software_ from Redhat in the form of Fedora Directory System (LDAP system that trounces OpenLDAP in many respects), Redhat Certificate Management System, and projects like FreeIPA. (which now means that Linux in the enterprise is starting to have a rival Active Directory)

Redhat purchases Jboss, and in response the Jboss becomes completely _Free_software_ moving away from their custom 'open source' license to things like GPL, LGPL, Apache, etc.

And Redhat spent a great deal of time, money, and effort, creating viable compiling Java implimentation and porting many applications over to it, which is probably one of the main reasons why Java is now moving towards Open Source/Free software.

etc etc etc.

Over and over again Redhat has a _proven_ track record of being willing to spend _millions_ of dollars purchasing significant corporations and open sourcing the software under Free software licenses. They have shown nothing but commitment towards Free and Open Source software that drives their support-based business model.

There is no doubt in my mind that whatever software related to SPICE that Redhat feels there is a need or a purpose for, they will open source. These sort of things are difficult as SPICE or other related software may have copyright owned by third parties or have have patent issues, but I have decent faith, built from historical evidence, that Redhat will open source what they can and not build anything for their customers that depends on closed source software.

------------------------------------------------------

This 'Oh what about in the context of Launchpad' hand waving is complete crap. It is utterly unrelated and as far as I can tell is just a excuse for you to tie in a favorite subject that you like to harp on.

It has _nothing_ to do with Redhat and it has _nothing_ to do with this article.

SolidIce and the SPICE protocol?

Posted Sep 12, 2008 6:21 UTC (Fri) by buchanmilne (guest, #42315) [Link]

Fedora Directory System (LDAP system that trounces OpenLDAP in many respects).

Uh, no (unless you are going to list those "respects", I suspect they are no longer valid).

SolidIce and the SPICE protocol?

Posted Sep 11, 2008 0:43 UTC (Thu) by SEJeff (guest, #51588) [Link] (4 responses)

I'm not fully agreeing or disagreeing with you Jef.

I've just seen you post very critical posts about Ubuntu in more than one
article on lwn. I don't need to prove to you what you know you are doing.

Ubuntu being successful == fedora being successful because it == Linux /
open source being successful.

While I agree 100% about launchpad being open source, mark shuttleworth
said it will be open source by next year's oscon.

http://arstechnica.com/journals/linux.ars/2008/07/23/mark...
launchpad-to-be-open-source-in-12-months

I guess I'm just tired to seeing free software / open source guys flame
eachother. There is no point, we aren't your "enemies". Ubuntu Hardy took
the xrandr gtk patch and the xrandr based display capplet from fedora while
fedora took upstart from ubuntu. Everyone wins when we stop bickering and
start working together. That is all.

Jeff "dyslexia spelled spaleta wrong" Schroeder

SolidIce and the SPICE protocol?

Posted Sep 11, 2008 7:32 UTC (Thu) by Lovechild (guest, #3592) [Link]

We have seen Mark claim opening Launchpad for years, fact remains that it is not and that since 2006 users of Ubuntu are forced to use non-free software to report bugs, as are packagers of software that uses Launchpad. This promise is worthless at best based on prior disappointments - what would work to dispel the critisism over Launchpad would be freeing it now and additionally henceforth in all projects adhere to the release early, release often credo.

In that sense I think it's entirely fair game to compare the track record of Canonical with the track record of Red Hat (or Novell for that matter, just to hit the big 3 distro vendors). Sadly it looks rather poor, I wish it didn't since Linux would be better as a whole if it wasn't the case.

Excuse some of us for critising Canonical based on what they do and not what they say. It does not mean we want to gleefully watch Ubuntu burn, but we do want their parent company to play nice with the other children. Especially given their highly vocal CEO's multiple statements on the record as to the virtues of open development and proposals as to how the rest of us should work.

SolidIce and the SPICE protocol?

Posted Sep 11, 2008 21:02 UTC (Thu) by jspaleta (subscriber, #50639) [Link] (2 responses)

Mark Shuttleworth has a tendency...to speak out of place. No one should take anything Mark Shuttleworth says at a Conference or in his blog or in an interview at face value. He's a very very good...wordsmith.

https://bugs.launchpad.net/launchpad/+bug/50699/comments/43

"N.B. - the roadmap has to be approved yet and some further detail
added to it but we're thinking about this seriously."

Now maybe you read the phrase 'thinking about it seriously' as meaning 'firm commitment to a specific timeline and roadmap' but I'm not sure the Ubuntu community members who have so far commented on that bugreport do.

The problem isn't Ubuntu as a community. The problem is Canonical and its desire to centrally control how the Ubuntu contributor community is allowed to grow and interact through the control of Launchpad.

The problems with just Lauchpad's Rosetta translation component makes a pretty good read as an impetus as to why the Ubuntu community needs to see Launchpad opened up so that the Ubuntu community can dig in and adapt the tool that Canonical has created to better serve the Ubuntu community's desire to work better with upstream projects.

2005:
https://lists.ubuntu.com/archives/rosetta-users/2005-Apri...

2006:
https://lists.ubuntu.com/archives/desktop-bugs/2006-April...

2007:
http://ubuntuforums.org/showpost.php?p=3676636&postco...
http://charismacode.blogspot.com/2007/01/powers-and-repos...

2008:
https://lists.ubuntu.com/archives/ubuntu-translators/2008...

Active, dedicated Ubuntu community members have been calling for open access to a critical piece of the infrastructure which they rely on to do they work they are doing. They have been doing that for far longer than I have been talking about it. I don't really care whose motivations you find compelling. Nor do I care if I'm cast as "the enemy" what matter is Launchpad as it stands right now is an impediment towards closer collaboration between the Ubuntu contributors and the upstream projects.
That is unlikely to change until Launchpad is open for community collaboration so that the Ubuntu community can adapt it to better meet the community's needs...regardless of what Canonical needs Launchpad to be as a profit center. If Canonical isn't going to respond to years of internal Ubuntu community pressure to open up Launchpad. Then perhaps they'll respond to external pressure by having an external party..perhaps even lwn...taking the time to compare and contrast exactly how dedicated to community and the open software ecosystem Canonical really is.

-jef"...next up Canonical's self-serving double standard with regard to Ubuntu's trademark when it comes to the Netbook remix"spaleta

SolidIce and the SPICE protocol?

Posted Sep 11, 2008 21:12 UTC (Thu) by SEJeff (guest, #51588) [Link] (1 responses)

Jef,

I agree with you completely. THis is kind of OT for this article though. Care to take a bet on how long before the SPICE protocol is fully documented with non-emcumbered OSS code out? I'd give it 2 years.

SolidIce and the SPICE protocol?

Posted Sep 11, 2008 21:44 UTC (Thu) by jspaleta (subscriber, #50639) [Link]

Difficult to know.

I think the the Netscape Directory codebase took longer than that to fully open..but that was much more than a protocal definition. That directory codebase release had to be staged, the gui admin tools came later because they essentially had to be written and couldn't be opened.

Advanced Message Queuing Protocol isn't a good example either because it was a from the ground up specification that Red Hat has been working on with partners.

The closest previous example maybe when Red Hat purchased Sistina in 2003 and then subsequently open sourced their GFS implementation about a year later. I wouldn't expect it to be faster than how long it took to get GFS opened.

But it really comes down to what the legal particulars of the codebase in question.

Of course none of that historical comparison should keep anyone from prodding Red Hat to do it as fast as possible.

-jef

SolidIce and the SPICE protocol?

Posted Sep 5, 2008 5:04 UTC (Fri) by dowdle (subscriber, #659) [Link]

See my expanded blog posting entitled:

Qumranet Joins Red Hat - Lots of questions
http://www.montanalinux.org/redhat-buys-qumranet.html

Your comments are encouraged.