OpenVZ Pushing for Linux Kernel Inclusion

RomanianClimber writes to tell us News.com is reporting that SWSoft is trying to get OpenVZ into the Linux kernel. OpenVZ is an operating system level server virtualization solution, built on Linux. From the article: "In this, it has a major ally: Red Hat, the top seller of the open-source operating system, which plans to add the software to its free Fedora version of Linux for enthusiasts. The companies' move to make OpenVZ partitioning standard in Linux is timely, said Pund-IT analyst Charles King."

Why is this needed?

You can compile anything you want into the kernel.

If this becomes part of the official kernel, then it becomes the kernel maintainer's problem.

If Red Hat comiles it into their distro's kernel, it is Red Hat's problem to maintain.

So if I were the kernel maintainer, I would need a very compelling reason to take on the extra work.

Re:Why is this needed?

The kernel maintainers have as policy that they won't give you a stable source interface, if you want your driver to work well you should get it into the kernel (See Documentation/stable_api_nonsense.txt). That's fair enough, but a policy like that gives them a responsibility to accept things into the kernel.

Re:Why is this needed?

1: hans?

2: Not just pay for but work with. This is the reason Xen has never really gotten into vanilla, even though it is supported directly by IBM, Intel and AMD.

Anyone worried?

Now, I've seen SW-Soft at work numerous reasons and I don't quite agree with their principles of development. Just check out their forums, they have an awesome community of people asking features in their higer end products and they never want to implement those. Instead, they're creating some kinds of "solution" to allow "lower TCO" and "easier management", at an extra cost of course. I've used their software, and it's quite buggy.

Now, Virtuozzo is one of their most awesome products, but I still don't feel right about having a company control over a piece of software embedded into a kernel. I have a chilly feeling about what they might do next and about what they're actually gaining by enabling this.

Just my two cents, I'm sure I'll get many replies of people disagreeing.

Re:Anyone worried?

I still don't feel right about having a company control over a piece of software embedded into a kernel.

If Linus merges it into his tree, then how are they the ones that control it? If Redhat merge it into their tree, then how are they the ones that control it?

The whole point of Free Software is that the user is the one in control, not whoever happens to hold the copyright.

You do realise that other companies have lots of code in the kernel already, don't you? This is by no means the first company to push to have their code included in the kernel. SGI contributed XFS. IBM contributed JFS. Namesys contributed ReiserFS and are currently pushing to have Reiser4 included.

I have a chilly feeling about what they might do next and about what they're actually gaining by enabling this.

Some pretty decent reasons for this off the top of my head are:

1. They have less maintenance work to do (no updates every time a new kernel comes out).
2. Less hassle for their users.
3. More testers/bugfixers.
4. They gain positive publicity.

Why the FUD?

That's the beauty of OSS

If you want something in there, then by god, put it in there. There's no huge patchwork system that affects everyone using linux when one company wants to change the underlying functioning. They can do it, and sell it if they can, while the rest of us can go happily on our way not using it.

Memory is like an...

Memory is like an orgasm. It's a lot better if you don't have to fake it. --Cray Seymore

Re:Memory is like an...

He didn't miss a comma, he just faked it.

Hardware support?

Both Intel and AMD are releasing CPUs which support OS partitioning in hardware this year (2006). Does the OpenVZ project support or have plans to support these hardware features?

OpenVZ/VServer versus hardware virutalization

OpenVZ doesn't care about processor virtualization features. OpenVZ (like VServer) is all about implementing a system like FreeBSD jails. In this model, there's only one kernel running, but different sets of processes are isolated from each other through operating system features. The separation applies to things like the 'ps' command and the /proc interface in general, as well as things like sockets and networking.

With OpenVZ/VServer, you can set up security and network separation so that certain processes will think of themselves as on 'internal-web-server', while others will think of themselves as 'external-web-server', and the two sets of processes would not be able to interact with each other in ways other than through the same kind of networking connections that they would use if they were on separate pieces of physical hardware.

Something like Xen or VMWare achieves this virtualization by simulating separate processors, memory, and I/O space hardware. OpenVZ/VServer doesn't incur this overhead, but does require much more significant modifications to the Linux kernel, as lots of system calls have to be modified to enforce the process group separation rules.

IP Rights secure on this?

Has there been a serious investigation of potential patent claims against OpenVZ. This looks like a potentially hazardous inclusion.

If due diligence has been done, and no problems on the horizon, then that's great. Just would hate to have something like this included and have it open up another SCO-like situation. Recognizing that one is Copyright based, and the other would probably be Patent, and in particular US patent based.

Re:IP Rights secure on this?

Dude, questions end with a "?", not a ".". It just looks bad and is bad English.

I can understand one thing

If they want to have aspects of OpenVZ added to the kernel to support the lower level functionality like they do for UML and XEN then really this artical isnt anything really worth talking about it just seems normal and i wouldnt see Tovalds knocking back such a request.

If its anything more then that then yeah, they'll get told to bugger off (and so they should).

BTW i didnt RTFA on this one :)

Xen

Wasn't redhat doing a major Xen push too? Fedora Core 5 will include xen host and guest kernels plus xen3, and from what i heard their putting a major effort into getting that usable too.

Never bet on a single horse i guess?

Or am i missing something and are OpenVZ and Xen very different products? (doesn't sound like it)

Upside of Xen seems to be the ability to run *bsd and other OS's in guest domains too, no mention of this in OpenVZ

Re:Xen

Or am i missing something and are OpenVZ and Xen very different products? (doesn't sound like it) Upside of Xen seems to be the ability to run *bsd and other OS's in guest domains too, no mention of this in OpenVZ

Essentially, Xen creates a new kernel for each virtual machine instance (or dom-u), while OpenVZ appears to use the same kernel instance for each virtual server. The latter approach seems to have benefits for performance and scalability, but if you discover a kernel bug in an OpenVZ server, all other instances are immediately susceptible, whereas with Xen, only the dom-u you are in is exploited (though if all instances are running the same kernel, you're up the creek). You'd generally need to be able to exploit the dom0 in order to affect all dom-u's.

Obviously, you're right about Xen supporting multiple OSes per instantiation versus OpenVZ.

This is an idea I like

I have often wanted a means to run 2 OS instances on one machine. I, personally, would like a way to run FC5 and winXP simultaneously. This definately seems like a step in the right direction.

Running Windows XP and Linux simultaneously

There are several ways to do this, with varying levels of stability and performance.

QEMU [bellard.free.fr] will run Linux, BSDs, and Windows, from either Windows or Linux.
Colinux [colinux.org] will run linux from Windows XP. I'm not sure what the latest Fedora Image for it is, but I run a 2.6 kernel based Gentoo build from XP frequently (for that nethack fix).

I'm not sure either is suitable, but i would recommend looking at them, as they are both interesting projects, if not immediately useful to you.

Re:This is an idea I like

You don't want a VPS.... but your hosting provider does, especially now that off the shelf hardware is so fast that under full load, if you divide the CPU by 10 or even 100 under lighter loads, your'e still I/O and network bound.

Perspective

SWSoft are the makers of Virtuozzo [wikipedia.org] a commercial product that allows hosting companies to offer Virtual Private Servers.

A rival technology is Xen [cam.ac.uk] from Cambridge University, which is free.

Galaxy

Nice to see some progress in the Linux arena. But neither the quoted article nor the OpenVZ web site list too many alternative solutions. Here is one from another world (non-unix): OpenVMS [hp.com] Galaxy by Digital (now HP). Galaxy is part of OpenVMS, since more than half a decade.

http://h71000.www7.hp.com/wizard/wiz_3191.html [hp.com] (check the date - 1999!)
http://www.s-and-b.ru/syshlp/vms_html/6512/6512pro .html [s-and-b.ru] (an early online documentation, hosted by on a non-Digital/HP system)
http://h71000.www7.hp.com/availability/index.html [hp.com] (Lots of information about High Availability/Disaster Tolerance)

"All the world's a stage" or was it "All the galaxy's a stage?"
http://scifi.about.com/library/weekly/aa022800b.ht m [about.com]

User Space Linux?

How does this benefit over current inclusion of User Space Linux? Does it allow other operating systems a la VMware? Is it platform-agnostic? Any info?

-molo

Pros and cons

So what are the pros and cons of Xen versus OpenVZ? My initial reaction is that Xen is the way to go because it is REALLY running different Linux instances. This is good because you can upgrade different instances to different OS versions. I know on those big Sun boxes with virtual environments its a pain in the butt because to upgrade the OS you have to upgrade a zillion applications at once to the new OS version which is a nightmare. But with real virtualisation with completely different kernels running you can upgrade one virtual instance to a completely different kernel without affecting hundreds of other apps running on the same machine.

Virtuozzo and OpenVZ

Although Virtuozzo is "built on top of OpenVZ", is Virtuozzo's kernel component a publicly available version of OpenVZ, built without using any proprietary patches or modules?

http://openvz.org/documentation/tech/virtuozzo [openvz.org] states "Differently from OpenVZ, Virtuozzo(TM) is developed and designed to run production workloads in 24×7 environments ..."

and goes on to list, among Virtuozzo's advantages over OpenVZ:

"Higher VPS density. Virtuozzo(TM) provides efficient memory and file sharing mechanisms enabling higher VPS density and better performance of VPSs.

"Improved Stability, Scalability, and Performance. Virtuozzo(TM) is designed to run 24×7 environments with production workloads on hosts with up-to 32 CPUs."

Why should Linux accept a kernel patch if (unlike Linux itself) it is not designed to run 24×7 environments with production workloads on hosts with up-to 32 CPUs?

this has nothing on Solaris Zones

well this will probably run multiple kernels, but probably means multiple times the work and the administration headaches, with Solaris Zones [blogspot.com] you share the kernel, but you only need to administer one core install of the OS.
 
A base install of Solaris in a zone, uses just 100MB of harddisk space. And on modern hardware takes less than 15 minutes per zone to install. Of course if you use the latest and greatest Solaris Express releases, you can use ZFS+Zones [blogspot.com] to cut the size of each zone down to 50MB of disk space, and zone creation time down to create a zone in 1 minute or less [sun.com]. You could also download and install brandz(Solaris patches that allows user to run Linux binaries in a Solaris Zone), and have even more choice. If you wish to debug your apps, you can use a stable dtrace and debug userland of both Solaris and Linux. And the Solaris kernel.

Arg!

Hmmm, that smell. That smelly smell that smells..... like..... we demand that we are not demanding you do as we demand cause we got clout on our side. Hrmphf.

Pfft

Real men use
# chroot /var/secure /vmlinux

non-graphical interface?

What I'd really like to know is if there will be some way for me to ssh in to my server and "press" the power button for a virtual machine and have it start up. Or, will it require that I be able to export my display before I can start it up? And would there be any way to remotely grab the console of an already active virtual machine?

Similar to...?

Is this similar to FreeBSD jails http://www.freebsddiary.org/jail.php [freebsddiary.org] and Solaris Zones http://www.sun.com/bigadmin/content/zones/ [sun.com] or is it something new?

If it is, it's a good thing to have, though all that "commercial firm pitching a free version of their product into baseline kernel" thing sounds a bit dodgy.

Pund-IT?

A Pund company name... awesome.

*waits for it*

xen is king

xen has so much support, and soon in-hardware support from both AMD and Intel. xen is the virtualisation layer of now and the future.

VPSes like OS/2?

For anyone who has used this. Is the relationship like the way OS/2 (2.0-3.0) used to run Windows?

idea...

....Hey, let's just put EVERYTHING into the kernel! User space? Bah!

Xen ?

Pardon my ignorance, but how is OpenVZ different from Xen? If there is little difference, why not just make a bigger push behind Xen rather than attempt to get their own IP into the kernel.

The Linux devs should reject it's inclusion

All the current x86 virtualization stuff is going to be out of date soon. It will be just adding kludges to the kernel to implement stuff that required by virtualization deficiencies in old x86 stuff. If you need those kludges, Xen should be enough.

This is because Intel and AMD are going to allow new and far more efficient ways of doing virtualization, with hardware assistance (lookup Intel Vanderpool and AMD Pacifica).

So, I don't see much long term gain for the effort for all the minuses.

You risk lower quality and increased maintenance costs. And you might also increase exposure to patent claims (but I bet IBM can smash anyone to pulp especially with virtualization patents).

You will still need developers to work on Vanderpool and Pacifica stuff, and I think you'd get better "bang for buck" with that (plus I think it will be a lot more fun).

This would be a REALLY REALLY bad choice.

I myself work on software which uses a VServer modification to the kernel. Although I do see advantages to setting this up so that it's included into the kernel. I see many more problems that this create then the good it would bring through.

Two really big problems I see are these two.

1) There is many other virtual server projects which do the same thing as OpenVZ. If one is included into the kernel, and the others conflict with eachtother over that, that's really going to complicate the linux world.
2) Multiple projects use vserver software currently in project, or they are developing on one of the many different virtual server project. This would cause problems for every one of those peoples project. Companies could loose lots of money because of a foolish decision like this.

The choice should be up to the user, and they should not be restricted to any one server virtualization project. This would get rid of competition over virtual server projects. If they are going to include this virtual server software, they should include all of the current virtual server projects and make them options. Most of them are probably incompatible with eachother, so the code has to make sure those conflicts do not happen.

Maybe an alternative should be to have a patchset made by the OpenVZ which could be given to linux for each kernel release, and multiple trees could be made. A regular kernel, then alternative virtual server kernels.

To allow this to happen would be something like Xorg saying they will only support Intel video cards from now on. Anyone with anything which doesn't have the intel chipset on their video card which is supported is screwed. Or for the linux kernel to only support AMD processors, it just wouldn't make sence. The foolish decision of OpenVZ to request this above all the other server virtualization projects is an extremely greedy and foolish choice I think.

I hope linus says no, or comes and checks the slashdot comments to read this and then tells them no. I may even have to fire him off an email about this.

While I can understand OpenVZ's side of things, overall this would be an extremely bad decision. I hope this never comes to be, for it will be a very sad day.

As for OpenVZ, Quit with the greed, keep your project as a seperate kernel addon to give a more competitive market.

5 (euro)cents worth...

My opinion? (as if anyone cares!)

This shouldn't be in the kernel.

-/Most end-users won't care about this technology. For them, Hypervisor is more of interest (whoa! no more dual booting, dude!!)
-/The functionality is more relevant to businesses who have no issue with custom kernels.....but then why not go for something supported/enterprise-grade?
-/It's the crappy freeware version intended to sell the upscale enterprise version. Why put something intentionally crippled in the kernel?
-/ There are several similar technologies being worked at. What makes this entry-version freeware so important?
Now, don't misunderstand me, I'm all for this type of software (even if not as full featured as the commercial package, I would still like to express a lot of gratitude for releasing this). I just think that the people who want to play around with this, will have no problem applying the patch to their kernel. I see NO reason to include this in the kernel.

Is it to get ready to GPLv3?

Earlier SWsoft was playing on a thin age of GPL conformity. Custom modifications to the kernel to enable proprietary kernel modules to work were "GPLed", the real stuff was not (as far as I know). I am sure, OpenVZ was born under the external pressure of the customers, but in the same time, the company was preparing patents, and will certainly use them to squeeze money. So, for SWsoft it might be important to push `free' code into mainstream kernel, before GPLv3 arrives in its full power.

VZ hosting products are interesting

E.g. http://globalservers.com/ [globalservers.com]

It is like colocation but less expensive. You share cpu & disk with other servers, but it is easier to maintain than a complete dedicated server.

OpenVZ versus Zen versus User-mode Linux versus...

There are so many different virtualization projects nowadays. And I keep hearing about magic new CPU features added to the latest Intel or AMD chips which will make virtualization easier than it is today, in some way. Is there a comparative review of the various approaches?

(I still miss running dosemu... that was fun...)

Re:Top seller

You do realize that when you buy Red Hat Enterprise, you're actually paying for technical support.

RHE is based on CentOS (http://www.centos.org/ [centos.org] It's also my favorite distro for both desktops and servers! So go try it out.

Re:Top seller

RHE is based on CentOS

You've got that backwards. CentOS takes the RHEL SRPMS released by Red Hat, rebuilds the binaries, and reassembles them into a distribution.

Re:OpenVZ?

Other posts have covered it, but a quick summary:

OpenVZ is a subset of a commercial product called Virtuozzo. It provides "virtual private server" functionality similar to FreeBSD jail() or Solaris Zones, including a private virtual network stack, private process space, and such, to each instance. However, it all runs on top of a single (specially modified) Linux kernel. Its advantages are in easy resource sharing among instances - since everything is running under one kernel, resource sharing (disk, memory) is made simpler. However, it has the disadvantage of less isolation - if the kernel crashes or is subverted, the entire system is at risk. Also, unlike with Xen, for example, you can only run Linux distributions (with the same kernel version). You cannot run other OSes (like NetBSD, FreeBSD, etc.).