Red Hat Wants Xen In Linux Kernel

DIY News writes "Red Hat is aggressively pushing to get Xen virtualization technology included in the Linux kernel as quickly as possible. This move comes as Microsoft is pushing its own virtualization products and recently relaxed some of its licensing requirements around Windows Server 2003 to facilitate more pervasive adoption and use of those technologies."

Comments

Xen into kernel

[b100dian]

What exactly does "virtualization technology included in the Linux kernel" means?
That you can run virtual machines with that kernel? that that kernel can be hosted into a virtual machine?

Or that you can install parallel kernels and run part of the ELF binaries on the other machine?..

Re:Xen into kernel

[Elitist_Phoenix]

From a link in TFA:
"Xen is a virtualization technology available for the Linux kernel that lets you enclose and test new upgrades as if running them in the existing environment but without the worries of disturbing the original system"

Re:Xen into kernel

[caseih]

Since you've received few answers to you actual question, getting xen into the kernel means the xen patches required to run the linux kernel in a xen hypervisor (both as a guest and a host) will be a part of the mainstream kernel and be able to be built trivially. RedHat ships 3 different kernels now with FC4. One is a normal kernel available in both smp and non-smp configurations. Then we have the XenU kernel, which is a kernel designed to boot in a guest xen session. The Xen0 kernel is the kernel that you'd actually boot on top of xen and use as your main OS.

Once the Xen0 kernel is running on top of xen, you have basically a normal linux kernel running that does all the hardware support. Then you load up Xen guest machines running the Xen0 kernel and these run their in their own virtual machines complete with their own disk images and linux distro. So xen doesn't really have anything to do with running elf binaries on the other machine. If you ran FreeBSD in the guest, it would run those binaries inside of that OS and that libc. When Xen 3.0 comes out, if you have the new intel or amd chips that support on-chip virtualization, then Windows XP can even run as a guest underneath the linux kernel-Xen0 host.

Re:Xen into kernel

[secureboot]

Don't let that comment fool you though. Xen is much, much more. What if you organization had 4 distinct sites they wanted to host on one server? Start up 4 virtual machines, and back up their running state from time to time. If one goes down, just restart it from your clean backup IN SECONDS. Better yet, do it automatically. At the same time, Xen enforces separation from the host OS that the virtual machines are running on, so you don't have to worry about it being compromised (well, not in any way anyone has been able to demonstrate or even postulate yet).

Re:Xen into kernel

[hey!]

easier management of computing resource, unless I miss my mark.

In rough terms:

Admin Cost = N * (H + S)
where N is the number of computers, H is the network and system hardware admin costs for a single machine, and S is the sys admin costs for the machine. Distributing:

Cost = NH + NS

This is a gross simplification, since we all know that complexity is not a linear function of network size, but it will do to be goign on with. Now we take NH and by virtualizing the machine it becomes simply H, so

Admin Cost (multiple virtual machines) = H + NS

Basically, I think it'll be common practice in the future to create virtual machines out of thin air by copying a config file or some directories on a machine with available bandwidth. If the cost of enough surplus hardware is less than (N-1)H, then wouldn't it be cheaper to virtualize?

Of course the complexity is that costs aren't linearly related to N, or for that matter constant in the size and class of machine you are managing. Which is another way of saying YMMV. I think there's clear application in many kinds of situations, for example in software development where we're constantly worried about the various combinations of software our work will have to coexist with. It'd be very convenient to be able to pull a certain system configuration out of a library and have it up in a few minutes, then trash it after a few hours of use. But it may have potential in production environments too.

Re:Xen into kernel

[MyHair]

The Xen0 kernel is the kernel that you'd actually boot on top of xen and use as your main OS.

Nitpicking the "main OS" wording: it's the host OS. I would think in a production server environment you'd keep this OS minimal and not do much in the Xen0 domain so you don't risk crashing or compromising the host environment. On the other hand, if it's a game box it would make sense to have the 3d video drivers in domain 0, and if it's a workstation it may or may not make sense to have the host OS run apps and have the user domains for testing purposes.

Once the Xen0 kernel is running on top of xen, you have basically a normal linux kernel running that does all the hardware support. Then you load up Xen guest machines running the Xen0 kernel and these run their in their own virtual machines complete with their own disk images and linux distro. (my bolding)

Typo: The guests are XenU (user) kernels which typically have no real device drivers and are therefore much smaller. Very well put, though. Note that you can use any block device as a disk image: a file, an LVM volume or even an actual hdd.

Re:Xen into kernel

[b100dian]

Thank you, caseih, for the time spent explaining. Thanks MyHair for clarifying:).

Now it makes more sense;)

Re:Xen into kernel

[timeOday]

When Xen 3.0 comes out, if you have the new intel or amd chips that support on-chip virtualization, then Windows XP can even run as a guest underneath the linux kernel-Xen0 host.

Does that new hardware support allow running Windows without having to modify it first? For the moment, it appears that no OS will work without modification and support for Windows might not be forthcoming:

In addition to Linux, members of Xen's user community have contributed or are working on ports to other operating systems such as NetBSD (Christian Limpach), FreeBSD (Kip Macy) and Plan 9 (Ron Minnich). A port of Windows XP was developed for an earlier version of Xen, but is not available for release due to licence restrictions.

Requiring hardware support seems like a big short-term disadvantage to me. It sounds like a feature that will take a long time to trickle down from expensive "server" processors to become widely available.

Re:Xen into kernel

[Anonymous Coward]

You mean, Xen will finally enable Linux to support the notion of Live Upgrade(TM), something Solaris has had for years now?

Re:Xen into kernel

[Anonymous Coward]

This is a gross simplification, since we all know that complexity is not a linear function of network size,

Indeed!!! In fact it's often an inverse relationship!

In the networks I've managed, the 20-machine networks were far far more labor intensive than the 600+ networks; because in the latter case you have to have appropriate automation tools in place.

And you're not the only guy to do this bad math with Virtual-PC or Xen. A network of 20 Virtual-PCs or 20 User-Mode-Linuxes takes about the same much work as a network of 20 physical PCs.

The real benefit in scaling up is consistant platforms -- and if you're consistant platform is Redhat-on-Xen, yes, that works great; but if your consistant platform is Redhat-on-any-fixed-server-model that works well.

Your ideas of "create virtual machines out of thin air by copying a config file or some directories on a machine with available bandwidth" are really not simpler than copying a disk image, which is what you'd do in a distributed netwrok.

In my opinion the nicest thing about Xen is that all those silly department that can get way too much hardware approved (8-CPU machine for finance that the CFO approved for his department) can now share with other departments without them really knowing it.

Re:Xen into kernel

[dTb]

Without hardware support it will only be possible to run modified OSs (such as Linux with appropriate patches) as is currently the case. With hardware support it will be possible to run unmodified OSs (such as Windows).

Re:Xen into kernel

[Anonymous Coward]

Read more carefully before you start condescending. He specifically separates the sysadmin costs from the hardware maintenance costs, and you're blending them together (the automation tools you mention are sysadmin, not hardware admin). 20 virtual machines have roughly the same sysadmin costs as 20 physical machines, but their hardware maintenance costs are significantly lower (i.e. fewer motherboard replacements).

Re:Xen into kernel

[SWroclawski]

If one goes down, just restart it from your clean backup IN SECONDS. Better yet, do it automatically.

Where does Xen let you do that?
AFAIK that's planned, but doesn't yet exist.

Re:Xen into kernel

[brunson]

Xen can migrate a running virtualization from one machine to another, snapshotting a running instance is trivial in comparison.

Re:Xen into kernel

[secureboot]

Check out the Xen User Manual. Look at the xm commands possible. You'll see one is save/restore. That is what implements the functionality I'm talking about.

Another responder mentioned live migration being much harder - he's right. He's also right when he says this is also done, with some really, extremely, very cool results. You can migrate a Quake server from machine a to machine b iff you assume some SAN or NAS in 60ms downtime, PRESERVING NETWORK CONNECTIONS. That means if you're remotely logged into a virtual machine, and it migrates from machine a to b, you won't get logged out of the virtual machine!

This is ridiculously cool technology. Take note everyone. It seems like 80% of comments here are incorrect, however. Read the papers on the Xen site if you want to know how paravirtualization actually works. It takes time, but if you're interested, it's worth it.

Re:Xen into kernel

[bigredradio]

If one goes down, just restart it from your clean backup IN SECONDS.

Why would a system just "go down" without hardware issues that virtualization would not correct? I understand there is the likelyhood that somone removes or corrupts a system file ( rm -rf /* ) but for production systems that is less likely than hardware failure. If you have a test environment where you are contantly blowing away a partition or system files by accident, it would be awesome. I just would not rely on this as a backup solution.

Re:Xen into kernel

[Anonymous Coward]

From the user manual link parent posted:

"The drawback of this approach is that it requires operating systems to be ported to run on Xen. Porting an OS to run on Xen is similar to supporting a new hardware platform"

Does NetBSD have it yet? :P

Re:Xen into kernel

[secureboot]

It's not a reliable backup solution - if there is a problem, you have to fix it, or you'll just have it again, even if the recovery time is 2 seconds.

First - virtualization CAN correct hardware issues in that you can just migrate the virtual machine off the hardware, which takes between 60ms and 3s, and preserves network connections. Machine A dies? No problem, move your webserver, preserving all network connections, to machine b.

So in that case, you've just implemented redundancy quite easily, for as many services as you like.

Secondly - someone does rm -rf /, you just recover saved state

You have an attacker - recover saved state, fix hole, come back online.

As long as the disk where the state is saved is backed up from time to time, you have a backup of the operation of an OS - not just the files it uses.

The other cool thing you can do here is run a bunch of OSes each running your service. For instance, you have an OBSD, FreeBSD, Linux, and Windows server each running your pages on apache. One goes down (open bsd security hole, like always), you just put in the super-secure windows VM automatically, and you're protected from the vulnerability.

Just use your imagination.

Re:Xen into kernel

[secureboot]

Do you even have to ask?

Re:Xen into kernel

[mikefe]

The other cool thing you can do here is run a bunch of OSes each running your service. For instance, you have an OBSD, FreeBSD, Linux, and Windows server each running your pages on apache. One goes down (open bsd security hole, like always), you just put in the super-secure windows VM automatically, and you're protected from the vulnerability.

Just use your imagination.

My imagination tells me that you have that backwards*.

.

.

.

.

.

.

.

.

.

.

.

.

* For the humor impaired, yes I know it was a joke. I am making a joke also.

Re:Xen into kernel

[MyHair]

Does that new hardware support allow running Windows without having to modify it first? For the moment, it appears that no OS will work without modification and support for Windows might not be forthcoming:

From what I've read, supposedly the virtualizing technology in the upcoming chips will allow unmodified OSes to run as virtual machines. However, I haven't seen discussion of how hard devices will be shared between uncooperative OSes. Current Windows products expect to own all devices, and I expect the virtualization tech will handle registers and memory mapping, but I wonder where the mechanism is to keep multiple instances of Windows (for example) from conflicting over USB or any arbitrary PCI, AGP or ISA peripheral. Is this handled by Xen (in this case), the processor or something we haven't seen yet, or do we have to disable the devices in guest OSes to prevent conflicts?

Requiring hardware support seems like a big short-term disadvantage to me. It sounds like a feature that will take a long time to trickle down from expensive "server" processors to become widely available.

Agreed. Hopefully we're wrong and it will make it to cheap chips fast, but outside of server use and hobbyists I'm not sure there's a viable virtualization market yet. I think it could be a killer work-from-home app, though: install your company's VPN client and preferred apps with company settings in a VM and fire it up to do work...never worry that changing apps or settings in your home VM will mess up your work VM or vice versa.

Re:Xen into kernel

[timeOday]

I think it could be a killer work-from-home app, though: install your company's VPN client and preferred apps with company settings in a VM and fire it up to do work...never worry that changing apps or settings in your home VM will mess up your work VM or vice versa.

I already do this with VMWare, in fact. It's nice because they're pretty restrictive about the configuration of machines that can hook up to the VPN, so I have a virtual disk set up with all their scanning and inventory crap installed inside it.

Re:Xen into kernel

[SWroclawski]

Migration, yes, but not monitoring with automatic migration AFAIK.

Re:Xen into kernel

[ckaminski]

Nevermind the fact that databases, web servers and software in general just IS NOT designed for point-in-time redos. Especially tiered applications. Apps will need to be rebuilt to take advantage of this feature.

Re:Xen into kernel

[ckaminski]

I already do this with Client setups whenever I start a new gig. Take my personal laptop which has greater and varied resources available to me to perform a job, and install the specialized environment they require to work on their network. I simply don't enjoy carrying two laptops around.

Hmmm

[dbgeek]

What a novel idea.

Xen...

[Anonymous Coward]

Xen isn't all that hard, you just need some jump boots and a particle weapon of some sort.

Re:Xen...

[Verteiron]

Be sure you don't fire on the workers in the alien factory, either, or they'll all turn on you.

Re:Xen...

[DrSkwid]

Don't worry, I shot them all dead!

Re:Xen...

[Jesus_666]

See, if Xen was already running inside vanilla Linux you could have just killalled them.

Umm

[interiot]

Well, Xen is free, and Intel/AMD hardware solutions are comming soon, which will allow Xen to run Windows unmodified. So, once everyone is upgraded to the new CPU's, virtualization will become a basic standard feature for everyone. MS can compete by giving their solution away for free, but either way, it doesn't get better than free for the consumer.

Re:Umm

[The OPTiCIAN]

Well.. Microsoft *could* try paying me to run their solution.

Re:Umm

[LaughingCoder]

I am having trouble understanding the rules of engagement. If MS "gives away" IE for free, they get called to task. If they "give away" Media Player for free, they get in hot water. Now why exactly would giving away virtualization not result in the same harsh treatment? Is it because there aren't established for-profit companies in that space already? How about VMWare? If MS gave away virtualization, the "anti-trust" crowd would drag them into court faster than you can say "billable hours".

Re:Umm

[DrSkwid]

They got in trouble for bundling both I.E. and media player.

They had to re-engineer Windows to make I.E. replacement simpler.

Re:Umm

[NitsujTPU]

Yes, that is my main reaction. Also, mind you, that virtualization integrated into the OS is completely unnecessary. Full virtualization (which is not OS dependent) really offers the fastest and best support for general use.

Xen has a number of interesting features that require paravirtualization techniques, which would need to be built into the OS.

Re:Umm

[top_down]

Giving things away for free is bad when it is used to stiffle competition.

There are basically 2 ways in which you can do that: by dumping or by creating private standards. Dumping is selling stuff below cost price (and thus taking losses) until your competitors are out of business. Private standards can be used to make competing related products incompatible or generally inconvience users of those products and thus try to set up a monopoly, this is at the core of Microsofts business strategy.

For many people this is all too abstract so they reduce it to 'giving away products is bad' which is clearly nonsense.

Re:Umm

['nother poster]

Microsoft got in trouble because the integrated their browser and media player into the operating system. If they would simply have had a browser and media player on your Windows install CD that you could install, or not, there wouldn't have been nearly the ruckus. What they did was make the install of the items required, and made it impossible for the average user to remove them without crippling the operating enviroment.

Operating system virtualization on the other hand pretty much requires it be hooked into th OS by its very nature.

Re:Umm

[Surt]

This got moderated funny, but actually this is a common business tactic, sometimes referred to as flooding the market. The idea is: put all your competition out of business by running at a loss for a period of time, then jack up the price as a monopoly and recoup your investment.

Re:Umm

[NutscrapeSucks]

There's a lot of misinformation on this thread. Microsoft actually wasn't convicted of "dumping" or "bundling", at least in the US. They were mainly busted for exploiting their contracts with OEMs.

Furthermore, all of their anti-trust problems were in the desktop market. As long as virtualization was positioned as a server feature, and as long as MS didn't threaten any VMWare supporters, I don't think they would have any legal problems.

Re:Umm

[div_2n]

MS has a monopoly on the desktop. When they start bundling software that does X when there are competitors that offer similar software, MS has abused their monopoly when they actively work to exclude those competitors and force a choice on consumers.

In other words, if MS decides to start shipping software to learn how to speak Italian with every copy of Windows XP, then they would have to allow makers of competing software the same courtesy to ship alongside their own.

If they didn't command a monopoly of the desktop (and weren't already convicted of abusing their monopoly) they wouldn't be bound by such requirements.

Re:Umm

[Tinik]

It's also illegal in the US

Re:Umm

[Malor]

That's only illegal if you're already a monopoly. It's illegal for Microsoft to do that, but not for, say, Red Hat. They're not supposed to leverage their monopoly position in one market to dominate another.

As a rather off-topic aside, it strikes me that Sony and Microsoft selling their game consoles at a loss is doing exactly that. Sony makes some profit now on their game division, but Microsoft has lost billions.

From this armchair, that sure looks like classic monopoly abuse.

Re:Umm

[erlenic]

It's really not applicable in the game case. Nintendo is the longest still surviving in this market, but (I think) Sony is doing the best (arguably.) They compete with each other quite a bit, so neither have monopolies, unless there's an early court case I'm unaware of. Microsoft, on the other hand, does have a monopoly, but not in the game market, and they're not leveraging that monopoly to knock everyone else out of the game market. If they made it impossible to use IE to lookup information online about the Sony and Nintendo consoles, then they'd possibly be illegally leveraging that monopoly. The closest they come right now is leveraging their money and name, which is legal.

Re:Umm

[Surt]

Oh sure. But it only impacts you if somebody enforces it. For example, Microsoft sold huge numbers of xboxes at a loss to try to gain market share in the console space, but has anybody gone to jail? Has Microsoft paid any fines?

Re:Umm

[N1XIM]

While the project has kinda been in extended hibernation while I figure out how to make it compile on newer kernels with newer GCC versions, Plex86 (the original, on savannah) would run unmodified guest OS code. It wasn't fast mind you, but it was/is elegant (and I hope that it will be again--Kevin Lawton & Co. wrote some really cool code that mere mortals like myself now have to figure out what to do with....). Now that I have a real job, and am working on putting together a development machine with a little more horsepower I hope to make it work again--and to make it faster.
The other unmodified guest hosting options out there are QEMU and Bochs (also by Kevin Lawton) both of I which have have also used at various times. They both use different technologies (and bochs was code-compatible with the original Plex86 for a while--until they made the UI too heavy....oh well).
I've been wanting to make a lightweight Plex86-based host for a while now, but that will probably go kinda slowly......

Re:Umm

[interiot]

Dude, do what I'm doing, and wait for the new CPUs from Intel or AMD. Xen WILL run Windows unmodified on these new chips once they're available, AND will execute slightly faster than VMware currently does (eg. like 95% of normal speed). See here for the Intel/AMD info. Or buy one of the Intel Macs once they're available, they're rumored to also use the Intel VT chips.

Re:Umm

[Al Dimond]

And yet look at KDE. Look at any end-user-centric Unixy distro (Gentoo, Debian and the BSDs don't count here). If you consider the windowing and desktop environments to be "part of the operating system", and most people do, then these situations are exactly the same. You can't have a KDE system without Konq. And yet we don't yell, "Evil KDE! Evil Knoppix!"

Actually I was recently installing FreeBSD on a laptop. My ISP responds to any request you make with an HTML page containing its terms of service until you click the "Accept" button at the bottom. After FreeBSD gets done installing (I used a CD burned by my desktop machine for the install; these days there are 2 install CDs, the first one required with no packages and the second one a bunch of optional packages. I don't keep a stockpile of CD-Rs, so I only burned the first disk and figured I'd just install everything else later by FTP. Well, you know where this is going. Installing everything on the "required" disk (which contains some optional packages like an X server, source code, documentation and "games", all of which I installed) doesn't give you any kind of web browser whatsoever. Not even Lynx. In fact, you don't even get wget. You get telnet, but I didn't really feel like looking up the HTTP spec and typing in stuff manually. Not to say I haven't done it before ;-). Eventually I had to d/l the Lynx package onto my PC and transfer it by floppy disk to the laptop, which meant I had to *find* a floppy disk, no small undertaking. Ah, but the terms of use page wanted to be over https, and default Lynx package doesn't support https (crypto export/import regs, SSLeay restrictions for USians because RSA licenses are required... not quite sure why other F/OS browsers aren't similarly affected, though SSLeay might be an odd cookie in this regard). So, grab the BSD package for Lynx w/SSL (which given the long page of possible issues at http://www.columbia.edu/~ariel/ssleay/ssleay-legal -faq.html is probably a violation of international law), it just *barely* fits on the floppy, finally get it working. Note that because of the https thing telnet with HTTP commands wouldn't have worked anyway. That makes my soul cry. I don't think you can just use SSH for that either, though; currect me if I'm wrong.

Somehow, I find myself thinking bundling wasn't such a bad idea (and yes, I know that bundling doesn't mean "do what MS did", this is just a silly story about the hassles of not having a web browser).

I thought what Microsoft got in trouble for was pressuring OEMs not to offer alternate operating systems or programs like Netscape with the threat of losing their Windows discounts or ability to ship Windows altogether. But I am not super-knowledgeable about this at all.

Re:Umm

[N1XIM]

I'll believe it when I see it.

Re:Umm

[TheRaven64]

Microsoft, on the other hand, does have a monopoly, but not in the game market, and they're not leveraging that monopoly to knock everyone else out of the game market.

It could be argued that they are. They have a monopoly on desktop operating systems, and through that a monopoly on desktop gaming APIs (DirectX - even OpenGL games tend to use DirectX for non-graphics things). By providing DirectX for the XBox, it could be argued that they are making it easier for game developers to release for Windows (a big platform) and XBox than for Windows and PSn, which might be an abuse of a monopoly position.

It's kind of tenuous, and it might not stand up in court, but there are arguments that can be presented.

Re:Umm

[interiot]

Consider these news items:

• Xensource gets $6,000,000 in funding
• AMD is actively working with Xensource to support their new chips

Also, not directly tied to running non-modified OS's:

• IBM contributes to Xen

The proof isn't in the bag quite yet, but there's quite a bit there to say that Xen will be really important in the near future.

Re:Umm

[N1XIM]

True, these are not minor developments in the world of XEN.
I, however also have a training in CPU design and can tell you that there is a whole hey of a lot more than meets the eye with all of this.

Re:Umm

[drinkypoo]

you still can't replace IE, it's an absolutely necessary component for providing embedded web browsing and HTML parsing to assloads of programs. Even Unreal Tournament uses it. There is a patcher that will cause these programs to embed mozilla (or there was anyway) but you can't expect good results when you do that.

Re:Umm

[Kethinov]

You know, I'm no fan of Microsoft, but this argument gets real old. If MS didn't include IE with Windows, exactly what would we use to download Firefox? Yes, it would be real super cool if MS made FF their default browser so as to be impartial, but the reality of the situation is there's nothing evil about MS making their own browser and setting it as the default.

The flaw with IE is that it's integrated into the OS and can't be uninstalled, yes, but not that it's included with Windows. Is Safari evil for being included with OS X? Is Konqueror evil for being included with KDE? Is Firefox evil for being included with Ubuntu? No, and for the very same reason IE is not evil for being included with Windows. It is only evil for being impossible to remove and totally integrated into the OS.

Re:Umm

[interiot]

and can tell you that there is a whole hey of a lot more than meets the eye with all of this

Um, yeah, you can say that again.

I don't quite understand how the new chips significantly reduce the complexity of vmware-style programs. I guess they don't have to dynamically rewrite the code and constantly keep adding breakpoints, does the removal of that make the VMM significantly easier to code? I guess so.

Also, any idea whether the new hardware chips will make x86 virtualization provably secure? I'd heard that the VMware method of virtualization was at least theoretically exploitable, allowing malicious code in one VM to break out and run code in a separate VM.

Re:Umm

[bperkins]

It's also important to note that since Microsoft was considered a monopoly they have to play by different rules, at least according to federal antitrust regulations. Companies that are in competitive markets don't have this restraint. RedHat doesn't have a monopoly on anything, so they can bundle all they want.

Adding to their trouble was that Microsoft didn't hide the fact that they were out to destroy their potential competitors very well.

Re:Umm

[N1XIM]

The main problem with x86 chips and true virtualization (taking one of something and pretending to have many of it) is that some rather commonly used instructions modify global properties of the processor.
What VMWare (and everybody else, for that matter) does is checks for those instructions and emulates them out of the bytestream executed on the real CPU. Some projects/products have tried for a "safe execution" mode, where all branches lead to known safe destinations (this is how the original Plex86 works--and why it can be quite slow at times), but I don't know if VMWare does this or not. For all I know, VMWare could just be encapsulating code run at each ring-level (ring0 is kernel mode, and has the world as its oyster--ring3 is user mode, and is mostly CPU bound) and only dealing with the "unvirtulizable" instructions commonly used in each ring-level (which would virtualize a hell of a lot faster than checking the for the whole set of unvirtualizable instructions, as you may have guessed)--and then catching hardware access attempts in some other manner.
In the case of Plex86, I know that what slowed it down most was the time it spent in userland emulating the hardware and pre-scanning the code for improper addresses (and chaning them) or touchy instructions (breaks, jumps, idles, state changes, etc.)--the kernelspace execution was reasonably zippy.
In any case, to overcome the need to pre-scan the code for globally reaching instructions one would effectively need to partition the CPU (in a true x86 implementation) and have some way of letting code run in one partition be aware of code run in the "slave" partition (and meanwhile also not letting the slave talk directly to the real hardware devices, mind you). I'm guessing that AMD/Intel are going to make a minor adjustment to their multi-core dies (and the firmware/microcode they run on....yes CPUs can have both firmware and microcode now) to allow this--and probably supply the outside world with access to this property via some sort of BIOS/ACPI endpoint/virtual device to talk to.

Re:Umm

[Trepalium]

There's nothing wrong with bundling. The problem was that Microsoft is a monopoly, and the use of bunding was to expand their monopoly into another area by giving their version of the product way under cost. If you assume your time has a particular cost associated with it, it was more expensive to download another free browser than it was to use the one built-in to MS Windows.

The end result is that despite the fact that dumping is an anti-competitive tactic, the fact that Microsoft could leverage their monopoly over desktop operating systems in such a way that nobody could compete with them, even if you gave your product away. As long as their product was sufficiently capable, they would prevail.

Now, Netscape deserved a fair bit of blame for this, too. Frankly, nothing significant really changed in Netscape versions past 3, and the 4.x series just plain stunk. Netscape accelerated their own decline by not pushing their software hard enough, fast enough. To make matters worse, IE 4 was actually fairly good, and IE 5 was better.

Re:Umm

[maxwell demon]

And yet look at KDE. Look at any end-user-centric Unixy distro (Gentoo, Debian and the BSDs don't count here). If you consider the windowing and desktop environments to be "part of the operating system", and most people do, then these situations are exactly the same. You can't have a KDE system without Konq. And yet we don't yell, "Evil KDE! Evil Knoppix!"

KDE hardly has a monopoly.
A happy GNOME user.

Re:Umm

[Anonymous Coward]

Microsoft got in trouble for doing all these things (and more) while also being ruled a "monopoly". Non-monopolies can resort to any kind of non-fraudulent behaviors, since it is assumed that the market place will sort it out the good competitors from the bad. Monopolies (it is suggested) disallow this remedy, because competitors who offer more consumer friendly behaviors aren't allowed to survive in the same waters.

No distribution that packages KDE is considered a monopoly by the U.S. (or any) government. Hence, they can all bundle things to their hearts content; indeed, they have to in order to be considered competitive in the marketplace.

Re:Umm

[interiot]

Intel released their machine-code-level spec a couple months ago. I'm not sure AMD has released theirs yet. But yeah, they just do a couple minimal changes. I haven't seen a good technical review of it yet. I understand the document generally, but I don't know if all the I's are dotted and the T's are crossed.

Re:Umm

[Al Dimond]

So the problem is a monopoly offering anything under cost? How are we separating the cost of operating system and browser here? Maybe they're gouging us for the browser and selling the OS under cost. We'll never know. But let's say that PC-BSD (www.pcbsd.org, I literally heard of it less than a minute ago, but it's an example of a Free Unix that "ships" with KDE installed) by virtue of its stability, ease of use, liberal licensing and unbeatable price becomes *the* dominant operating system sometime in teh distant future. Will it suddenly run into legal problems because it bundles a web browser, an office suite (probably has KOffice though I'm not positive), a development environment, and lots of other stuff, all for free?

Re:Umm

[Anonymous Coward]

You could do what I do: Use ftp.exe to ftp firefox from ftp.mozilla.org

I won't TOUCH IE on a new system installation. Once I get FireFox, I'll get AdAware and block of IE. That pretty much prevents spyware and bullshit from getting into the system.

Re:Umm

['nother poster]

Which is what I said.

Re:Umm

[dbIII]

If MS "gives away" IE for free, they get called to task.

Among other things, those that developed the application IE was based on didn't get paid becuase they were getting a percentage of zero dollars per unit. Spyglass were truly screwed over on that - then as IE improved it was used to clobber Netscape, after which it was almost abandonware for at least two years.

Re:Umm

[emilper]

Now why exactly would giving away virtualization not result in the same harsh treatment?

because Microsoft could take XEN and use it without more restrictions than other companies that want to use XEN for their OSes. Microsoft could even take over the development of XEN, or make a fork, if Microsoft chose to use GPLed software.

Re:Umm

[Trepalium]

No, the problem is the monopoly leveraging their monopoly to gain an unfair advantage over the competition. Dumping, and putting products on the market far below cost to drive competition out of business is already illegal (monopoly or not). Combining the two tactics only makes things worse (and more effective).

I'm sure that Open Source and Free software projects will be have anti-trust lawsuits filed against them in the future. Perhaps if some company actually gets big enough, they may even be successful. If it happens, it'll be one of the most interesting court cases to be tested against antitrust lawsuits in a very long time.

Two birds, one stone

[bernywork]

Not only do they get the ability to knock the shit out of Microsoft, by taking away the base platform from them, they also get to try to get some market share from VMWare.

Imagine if you would the ability to use Xen for unlimited operating systems, no licensing cost of the base OS, thinking about it, I would prefer to be in Microsoft's shoes as opposed to VMWare's. Only difference is that Xen when compared to VMWare is a very immature platform and no IT manager is going to take Xen over VMWare just yet (Unless cost is a BIG factor).

I would have to say that this is still very cool, with all the new Virtualization options come out in the new cores shortly and if they can get to market before Microsoft, this is a great way to pick up some customers. Kudos to RedHat and IBM and Intel and everyone else for making this happen.

Re:Two birds, one stone

[b100dian]

no IT manager is going to take Xen over VMWare just yet (Unless cost is a BIG factor)
With VM Player and vi you may get a free virtual machine.. see http://b100dian.lx.ro/wordpress/index.php?p=90

Re:Two birds, one stone

[Anonymous Coward]

I've been looking at XEN for a while now, if it were a configure option at my next kernel build then I guess I'd already be running it. VMWare is a different beast, horses for courses but XEN makes more sense than VMWare if required OS's have been ported.

Re:Two birds, one stone

[dc2447]

That gives the impression that Vmare is a great product, I'd debate that. I find it a PITA alot of the time. XEN isn't the only kid on the block, Qemu works really well too.

Re:Two birds, one stone

[Anonymous Coward]

You cannot compare Xen to the traditional VMWare workstation, because Xen runs at ring 0. Therefore, you should rather compare it to VMware GSX - Which IS an expensive product.

Re:Two birds, one stone

[DrWhizBang]

Not only do they get the ability to knock the shit out of Microsoft, by taking away the base platform from them, they also get to try to get some market share from VMWare.

Have you looked at the instructions to run something with XEN? I don't think this is a threat to VMware, in fact it is probably good for them because they can use it and it will make their product look better (better performance, better integration) and they can continue to sell an application to configure the virtualization system.

Re:Two birds, one stone

[Anonymous Coward]

You must have never used vmware before for it's best use.

vmware in linux is the absolute #1 tool useful for reverse engineering. I can pipe the rs232 and usb as well as ethernet ports fo the hosted OS to files or through sniffers on my linux machine and figure out quite quickly how a device is talking to it's hardware or server on the net. Capturing an entire session into a input and an output file makes is trivial to reverse engineer something.

this can not be done with zen or the utter crap that msft is making.

Re:Two birds, one stone

[bernywork]

I was more thinking along the lines of ESX server which contains a cut down OS for the base operating system, everything else (All other OSs) run over the top.

Re:Two birds, one stone

[squiggleslash]

Can we start prefixing acronyms, initials, and other re-used names, with the date they were coined or something? I'm reading this entire story and finding it hard to not think about ancient computers made by Apricot in the UK. Now you're blathering about CP/M's graphics standard, GSX? Argh.

Re:Two birds, one stone

[bernywork]

At the same time, how long do you think the tools are going to come out to manage all of this?

I truly don't think it will be that long. The hardware vendors IBM / HP etc will be willing to support this as if the Xen community comes out with extensions for the Virtualization stuff in the new CPU cores (also coming out shortly), then it will help them to push more kit (sell more hardware).

Re:Two birds, one stone

[j-pimp]

Imagine if you would the ability to use Xen for unlimited operating systems, no licensing cost of the base OS, thinking about it, I would prefer to be in Microsoft's shoes as opposed to VMWare's. Only difference is that Xen when compared to VMWare is a very immature platform and no IT manager is going to take Xen over VMWare just yet (Unless cost is a BIG factor).


I've been using xen here in what I call "production development." Its serving several development servers. One of them is running a crappy spam assassing frontend thats pretty resource intensive. I was quite satisfied when my colleuge asked me when I was going to move said test server from real hardware to Xen a week after I did the migration.

Anyway, while vmware is definatly prettier, I think xen performs better. Of course being xen can't run windows I can't see how one of the machines catching a virus affects the performance of all the machines. But if your comfortable with the command line, and have a decent grasp of lvm and piping tar over ssh, you should have no problem using xen.

Re:Two birds, one stone

[kv9]

free as in beer. and as other people have pointed out, you cant really beat the performance that the xen approach gets.

Re:Two birds, one stone

[laptop006]

But not the PCI bus, I'm probably going to have to use the PCI proxy code in mac-on-linux to do that (trying to implement an alsa driver for a pro soundcard without
  OS X drivers)

Re:Two birds, one stone

[BoldAndBusted]

And qemu works without any kernel changes. I just installed 0.7.2 last night, and I was amazed at how it could convert and run my VMWare Workstation *5* raw disk images.

http://fabrice.bellard.free.fr/qemu/

Erm ... Competition from VMWare Player?

[Anonymous Coward]

Does it have something to do with the release of VMWare Player? :S

Re:Erm ... Competition from VMWare Player?

[mrjatsun]

Yes. Why do you think VMWare released the free player? Competition from Xen.. ;-)

Re:Erm ... Competition from VMWare Player?

[LnxAddct]

Red Hat has been shipping Xen enabled kernels for months now in Fedora and I believe they are available for RHEL too. They have had quite a few people dedicated to working on Xen for some time, and that is a *good* thing . Virtualization is the future, and its good that a big company like Red Hat is pushing it further. Xen and SELinux are two killer technologies that Red Hat has really made viable, so kudos to them for keeping open source on the forefront of innovative (or at least uncommon, but much needed) technologies. For a truncated list of other cool technologies Red Hat is pushing in OSS, check out the Fedora projects page, you'll see virtualization, and SELinux in addition to the directory server, stateless linux, and system tap (our answer to Solaris' dtrace, granted it still is under heavy development and is far from ready for prime time)
Regards,
Steve

Aggressively pushing?

[gringer]

"My goal is to get this done in the most collaborative way possible with anyone in the community who wants to participate," Stevens said, adding that Red Hat is committed to putting on this project enough of its staff who have the technical knowledge necessary to get the work done.

Perhaps it's only me, but this doesn't sound aggressive; this sounds friendly and cooperative.

Re:Aggressively pushing?

[Anonymous Coward]

When someone intends to do something "in the most collaborative way possible", then it's not uncommon to find a silent "but I'll do it without cooperation if necessary".

Re:Aggressively pushing?

[bcmm]

I don't think it meant "aggressive towards the open source community". It's Microsoft they'll be competing with, and it seems that it's going to be Linux, rather than just Red Hat, against them.

So, they're "aggressively" pushing Linux instead of Windows as a virtualisation host OS. Six staff members hired to work solely on integrating it into the mainstream kernel is fairly aggressive (toward MS), I would say, as it could lose them a major new market.

Re:Aggressively pushing?

[Anonymous Coward]

I really don't care how many engineers the project has, how much money Redhat wants to throw at it, or whether or not Redhat wants to collaborate with others. All I care about is whether the technology is sound and well implemented. Putting politics in front of pragmatism is aggressive. Don't doubt for a minute that there are Redhat shareholders who would like the linux kernel agenda to support their bottom line rather than any technological consideration.

Re:Aggressively pushing?

[Anonymous Coward]

I think they mean 'agressive' in the business sense. i.e. "lets get this done quickly", rather than "lets schedule a pre-meeting for the spring to brainstorm some discussion points for a meeting in the summer".

Re:Aggressively pushing?

[caluml]

Take your rejected story to digg.com, and technocrat.net - they don't reject worthwhile stories.

Re:Aggressively pushing?

[bcmm]

It'd strange, actually. They covered previous releases, and even did several articles about planned features of this release before it came out.

Forking?

[Lardmonster]

Why don't they fork? Or just build and rpm their own kernel, like they did with GCC 2.96?

Re:Forking?

[LordNimon]

Because forking is more work than not forking.

Re:Forking?

[Jonny_eh]

I'm pretty sure they do that, most distros tweak the kernel to suit them somewhat.

I know that SUSE 9.3+ has built-in support for Xen, and since there are different kernel packages especially for Xen support, I assume that Suse has accomplished what Red Hat is working towards. Although, I could be completely wrong since I didn't RTFA.

Re:Forking?

[m50d]

Because that's bad for the community, something redhat cares about, even if the kernel developers don't seem to. They probably will ship their own xen version for the moment, but the less difference between their kernel and the mainline one the better it is for everyone.

Re:Forking?

[kwalker]

Red Hat (beginning in Fedora Core 4) has Xen RPMs for their kernels as well. Each time they release a new kernel, there are Xen versions as well. They want to get it into Linus' kernel tree so everyone can use and test it, since other distros are leery of using Red Hat stuff but fine with anything Linus has integrated into his kernel.

Re:Forking?

[VENONA]

I think Lardmonster was joking. The RH/gcc-2.96 fiasco caused a lot of problems. Like not being able to reliably run binaries built in this environment on other distros. A PITA for admins at large installations.

Re:Forking?

[sjames]

There is no need to fork in this case. The community seems to generally like the idea, so well done patches will probably be accepted. While most distros do maintain a set of distro specific kernel patches, they mostly come from one of the developer's trees, or are backports of generally useful features.

There are a few good reasons to submit a feature for inclusion in the vanilla kernel including reduced local workload for future versions. If the feature is mainstream, other developers will avoid breaking it unnecessarily while if it is just a local fork, they may not even know they're breaking it, and (correctly) won't be so concerned about it anyway.

A good reason to maintain a patchset for a local fork is a feature you need but that is not generally wanted or needed (the more invasive the patch, the more generally useful and well tested it will need to be for mainstream inclusion) or is considered too immature or unstable for inclusion. Another reason is to gain a useful new feature when you feel that the cutting edge kernel that introduces it is not ready for primetime. Redhat had a few such patches against 2.4.x to backport 2.5/2.6 features.

The irony

[Alioth]

The irony is that Microsoft provided some of the funding for Xen (probably for the early experimental Xenised versions of Windows XP). Yes - Microsoft does fund GPLd projects. Often in a company that big, the left hand doesn't know what the right hand is doing, so whilst Gates/Balmer spout off about how evil open source is, another part of MS is funding it (or even releasing it on Sourceforge).

Re:The irony

[drsmithy]

[...] so whilst Gates/Balmer spout off about how evil open source is [...]

No, both are quite specific it's the *GPL* they don't like, not Open Source in general.

Re:The irony

[photon317]

For the exact same reasons that I prefer the GPL to every other open source license. If you release code under a BSD-ish license, Microsoft can co-opt your work into a proprietary product directly without playing the same open source game that you are.

Re:The irony

[Anonymous Coward]

I don't think this is true.
Microsoft did fund some guy in the same lab, but on a different project - someone told me once.

Re:The irony

[Anonymous Coward]

No, both are quite specific it's the *GPL* they don't like, not Open Source in general.

Yes.. for now. This is called "divide and conquer" (MS playbook, pg.2)

Re:The irony

[Senzei]

For the exact same reasons that I prefer the GPL to every other open source license. If you release code under a BSD-ish license, Microsoft can co-opt your work into a proprietary product directly without playing the same open source game that you are.

At which point any updates to the code become their maintenance headache. With a BSD style license anyone that uses the code in a closed source project either ends up A) forking it and maintaining their own version or B) submitting their updates as patches. If someone takes BSD-licensed code for a project and forks it to add in their own code, they then have to explain to every PHB that hears about it why their code doesn't have features X, Y, and Z just like the new version of that project. In other words all "co-opting" code under a BSD license gives you is a jump start and a maintenance/marketing headache.

Don't get me wrong, I have no problem with the GPL, but there are things besides legal reasons why a license can effectively keep software free.

Re:The irony

[idlake]

At which point any updates to the code become their maintenance headache.

Yes, for most companies, that's a valid reason not to create a proprietary version; but Microsoft is big enough not to have to worry about that sort of thing.

Re:The irony

[AJWM]

If you release code under a BSD-ish license, Microsoft can co-opt your work into a proprietary product directly without playing the same open source game that you are.

At which point any updates to the code become their maintenance headache.

No, that just simplifies Microsoft's whole embrace and extend (and extinguish) normal operating procedure. The updates make the BSD code no longer Microsoft compatible.

For other companies, you may have a point, but the ancestor was specifically referring to Mircosoft.

Re:The irony

[SirTalon42]

Where do you think Microsoft got their network stack? How much code have they given back?

Re:The irony

[Anonymous Coward]

They like the idea of cherry picking other people's code for 'new' ideas, they DON'T like the idea of giving BACK stuff for free, unless it's total dinosaur code, and even then they get budgy. Let's face it, MS is STILL fighting Dr-DOS, almost a decade after DOS was shelved.

Re:The irony

[Anonymous Coward]

So modify the liscence. Add a clause to specifically deny Microsoft the rights to your code.

Re:The irony

[Anonymous Coward]

If you release code under a BSD-ish license, Microsoft can co-opt your work into a proprietary product directly without playing the same open source game that you are.

Freedom is not worth having if it does not connote freedom to err. It passes my comprehension how human beings, be they ever so experienced and able, can delight in depriving other human beings of that precious right.
        --Ghandi, 1931

Re:The irony

[Alioth]

Xen is GPLd. Microsoft contributed to the funding of Xen. Therefore, my original comment still stands - MS were funding GPLd software at the same time as whining about the GPL.

Re:The irony

[drsmithy]

For the exact same reasons that I prefer the GPL to every other open source license. If you release code under a BSD-ish license, Microsoft can co-opt your work into a proprietary product directly without playing the same open source game that you are.

There are plenty of other Open Source licenses that avoid this. The LGPL, for example.

Re:The irony

[Anonymous Coward]

Any now you get into the core definition of open source. In my book, open source is giving away code for ANYONE to use and thus the BSD License fits. GPL proponents believe that source isn't open but rather visible.. you can see it but you have to give me all YOUR code so that I may have it. What you try to protect, takes away rights from the developer that enhances it. Of course, stallman and the like say its the freedom for the software and developers can go to hell.

Microsoft uses FreeBSD for hotmail and therefore gives back by using it. They admit that windows can't run hotmail. This helps open source.

It means

[Anonymous Coward]

in part making Linux aware of the virtual architecture provide by Xen, not by Intel or AMD. Some of it is for better performance. Some of it is to make it just work. The latter is more worrying since there may be serious security issues as one of the major advantages of VM should be better security than that which is provided by Linux to begin with.

The other problem here is there are other VM's out there and they all have different requirements for kernel modifications so talk about mess.

The major underlying problem there is Intel and AMD just don't get it. They put in some rather pathetic and lame support for virtualization and think they're all done with it. They have this Not Invented Here attitude that has kept them ignorant of virtualization technology that has been around for decades and doesn't have as much of these problems.

Re:It means

[Octorian]

I still remeber reading that the whole x86 architecture didn't meet the requirements for virtualization, meaning that this recent trend is probably the result of VMware figuring out some "tricks to make it work", and then everyone else jumping on the bandwagon.

In any case, if you really want to learn about the fundamental concepts behind virtualization, I strongly recommend reading the following paper: Formal Requirements for Virtualizable Third Generation Architectures

Yes, it was published in 1974, but most of the concepts are still very applicable and make a lot of sense. (though the architecture examples are obviously dated)

This is a very good paper which lays out all the ground rules. Sure, it may sound a bit academic in terminology and explanation, but it is still quite readable.

Re:It means

[Viol8]

I was under the impression that both Intel and AMD were/are going to
add some new CPU instructions and another processor run level that
would support virtualisation at the hardware level (which needs to be
the case to do it properly).

Re:It means

[Chirs]

The full x86 architecture is not suitable for virtualization, because there are a few instructions which fail silently when run from user level.

VMware uses various techniques to get around this, including full simulation and binary re-writing.

Xen uses another approach, where they port to an instruction set that is basically x86 without the problematic instructions. This approach requires that the guest OS's be modfied.

This will all change with the new virtualization instructions being added by both AMD and Intel. Once that is in place, Xen will be able to run unmodified guest OS's (such as Windows, for instance). There will be a speed hit though, so modified guests will be prefered if speed is an issue.

Re:It means

[strstrep]

Wait... I thought this has been in these CPUs since the 80286! Ever used the virtual-8086 mode?

... /me ducks

Re:It means

[The_K4]

Here's a link for Intel's Virtualization Technology .

Re:It means

[Viol8]

Virtualisation is not the same as virtual memory or emulation
of an earlier CPU.

Re:It means

[kma]

VMware uses various techniques to get around this, including full simulation and binary re-writing.

I know I'm coming to this a little late, but the "binary re-writing" part is Xen/hardware vendor FUD. We translate supervisor-level x86 binaries into user-level x86 binaries, dynamically, on-demand, and adaptively. I.e., VMware's VMM never just branches willy-nilly into the guest OS's kernel code; it dynamically produces an unprivileged binary that has the same effect as the supervisor-level code, only on a software model of the CPU. Saying we "rewrite binaries" makes it sound like we either do so on the guest filesystem, or smash loaded in-memory code. Neither of these happens. Self-inspecting and -modifying privileged code works just fine in the VMware VMM. And "full simulation" is just plain misleading.

(Disclaimer: I work on VMware's VMM.)

Re:It means

[sjames]

Perhaps Afterburning will help. The original source is re-compiled with a modified toolchain that makes it easy to run the binary (such as a Linux kernel) unmodified on raw hardware or to fix it up to run cleanly in Xen or L4 (amongst others). The hardware changes will still be needed for proprietary OS like Windows, but would be a big help for any Free OS on older hardware.

Re:It means

[d1rty_d0gg_]

VMware uses various techniques to get around this, including full simulation and binary re-writing.

VMWare does not do a *full* simulation ala plex86 or bochs, instead it runs the guest OS at PL1 and translates only the sensitive instructions when they trap to PL0. Also note that the move to Vanderpool or Pacifica will benefit VMware as much as it does Xen - of course Xen may always hold the edge in terms of speed. That said, Xen's paravirtualization approach is quite kludgy - one can see how development issues will just go back and forth between OS writers and the virtualization people.

My Bias

[Vodak]

The anti Redhat Linux part of me is saying Do not cave into the demands of Redhat because they are becoming as bad as Microsoft with pushing Linux to their own sinister goals. But then the sane part of me says " If the technology is awesome it should be in the standard kernel."

And then the crazy part of me says. "Heh, I can compile modules for the Xbox controller and other weird hardware into the kernel. Maybe useful technology should be in the kernel =]

but then again. I just might have to many voices in my head

Re:My Bias

[RotateLeftByte]

Remember that anything RedHat pushes into the Linux Kernel will automatically become available for ALL OTHER LINUX DISTROS. So please forgive my ignorance, but where is the "badness" in that aim? AFAIK, nothing that RedHat has developed has been proprietary in any way. They do have a track record of buying things from other companies and releasing them as OSS stuff. Again please let me know of the badness in that aim.

IMHO, virtualisation is going to become very important to all sofware developers over the next few years. If it is easy to fire up a Debian system on top of a SUSE and have Mandriva & RedHat running as well then you can test your app on all these platforms at the same time. Hurrah!

Re:My Bias

[Erwos]

No, the sane part of you should be saying "Red Hat is nothing like Microsoft." So far, their own goals have been anything but sinister, and every other distro on the market has benefitted from the time and money they've invested in gcc, the kernel, and any other number of projects.

They've done nothing anti-community since dropping free Red Hat 9 support years ago. Get over it.

-Erwos

Re:My Bias

[Anonymous Coward]

"The anti Redhat Linux part of me is saying Do not cave into the demands of Redhat because they are becoming as bad as Microsoft with pushing Linux to their own sinister goals."

That's the retarded part of you. Learn to ignore it.

Re:My Bias

[ShieldW0lf]

Remember that anything RedHat pushes into the Linux Kernel will automatically become available for ALL OTHER LINUX DISTROS. So please forgive my ignorance, but where is the "badness" in that aim?

Because if you're not using it, it shouldn't be there? I can see why Red Hat would want it in the kernal, they sell enterprise server software, but with linux being used for everything from appliances to cell phones to PVRs to workstations to servers, seems to me there are LOTS of places where it would be undesirable.

Re:My Bias

[The Bubble]

But then the sane part of me says " If the technology is awesome it should be in the standard kernel."

Why? just because some piece of software is "awesome" doesn't make it potential kernel code. The kernel is getting too big and complex as it is... should the next move be to include X as part of the kernel?

:looks around expectantly for a Tanenbaum...

:wq

Re:My Bias

[Vodak]

I will freely admit that is it silly to dislike RedHat. And that for the most part the reason alot of people dislike RedHat is simple because they are the biggest and most known of the Linux Distros.

Have they done anything sinister? Not yet. Will they? Who knows. But it's fun to complain about them =]

Re:My Bias

[Vodak]

No; The retarded part of me is saying "RED IS PRETTY!!!" while eating crayons.

Re:My Bias

[justsomebody]

What a stupid argument

Because if you're not using it, it shouldn't be there?

So, basicaly what you are saying is "WFT is this rtl ethernet driver doing in kernel, I don't have this fscking network card! Anyone who does should apply patch"?

Typical computer has one CPU, one chipset, one eth... Now, just why do you think kernel contains more than one. Because people don't use the same one hardware you use maybe? Same goes for the needs, you have your own needs just as everybody else has their own. Personally, I can only benefit from Xen.

If module that is not desired by you is there, then that module can be compiled by choice. If you build your own kernel then nothing stops you to say NO to options you don't need. Beside that, I doubt that RH plans to enforce Xen by default.

I can see why Red Hat would want it in the kernal, they sell enterprise server software, but with linux being used for everything from appliances to cell phones to PVRs to workstations to servers, seems to me there are LOTS of places where it would be undesirable.

All those appliances build their custom kernel. And all will probably choose NO if they don't see the need for Xen. Where is the problem?

On the other side, all those who build those kernels can only benefit from this. They imidietly get more robust and better environment to actualy do some work on systems you named.

Re:My Bias

[Anonymous Coward]

You Sir are nothing but a troll. Its not fun to criticise someone who does not deserve it.
If I had some mod points that is what you would get modded as.
If you have nothing to say, then STFU.

Re:My Bias

[CrosseyedPainless]

ROFLCOPTER!

Re:My Bias

[mikefe]

You Sir are nothing but a troll. Its not fun to criticise someone who does not deserve it.
If I had some mod points that is what you would get modded as.
If you have nothing to say, then STFU.

Yes, sometimes it is ok to agree with an AC.

(c) me, 2005

[Library Spoff]

*hmmm*
Must remember to patent the idea of trojan/virus that uses visualisation to run a spam/DOS server
in a Windows environment...

Rubs hands with glee as he tries to sell the idea on IRC.

Re:(c) me, 2005

[Anonymous Coward]

"Xen and the art of kernel maintenance."

I hereby copyright the above sentence and assign the copyright to the EFF, please make all checks payable to the EFF. Now who wants to print the t-shirts? Virtually everyone will want one.

Re:(c) me, 2005

[digitalhermit]

Been done... The DoS is called Java.

(Ducking and running)

Just kidding. Java is my friend.

I see no problem here...

[bogaboga]

...If RedHat wants XEN in the [Linux] Kernel, they can put it in there themselves or they could pay someone to do it...or they could fork the Kernel. I have failed to see what is preventing RedHat from putting XEN in there. So right ahead RedHat. Go!

Re:I see no problem here...

[jurt1235]

Yes, that was the idea about OpenSource: If you don't like it, you can change it.

Re:I see no problem here...

[Milican]

RTFA

"Part of the Red Hat emerging technology team's efforts will be to drive the Xen virtualization technologies as part of the Linux kernel rather than as part of a sidebar project, as is currently the case, Stevens said."

JOhn

Sun could be in RedHat's crosshairs

[Zugot]

Sun can do this now with Solaris 10. Virtualization is a cool technology, and everyone in this space seems to be heading there.

Re:Sun could be in RedHat's crosshairs

[jdowland]

Well, sort-of, yes. Solaris 10's zones are in the same ball-park, but it isn't the same take on virtualisation as xen.

Re:Sun could be in RedHat's crosshairs

[Octorian]

Solaris "zones" technically aren't really virtualization, per se. Rather, they are virtual-machine-"like" process containers. Inside of a zone, it behaves very much like a virtual machine, but it really isn't.

This concept likely provides many advantages for system resource management on a server, where you only care about a single operating system. It does not, however, let you run different OSs at the same time.

Why Xen and not vservers?

[ptaff]

While Xen appears as a neat package, why choose Xen instead of vservers?

The hardware cost of running multiple copies of the same OS with vservers is smaller than Xen - there is one and only one copy of glibc in memory, one and only scheduler, and so on.

Re:Why Xen and not vservers?

[Anonymous Coward]

"While Xen appears as a neat package, why choose Xen instead of vservers?

The hardware cost of running multiple copies of the same OS with vservers is smaller than Xen - there is one and only one copy of glibc in memory, one and only scheduler, and so on."

But part of the purpose of a virtual machine is that you can run a different operating system in each partition, including different schedulers and libc versions.

Re:Why Xen and not vservers?

[Anonymous Coward]

With Xen, a kernel panic effects only that kernel, the other kernels keep on running. Under vservers, it takes down the machine.

Under Xen you can reduce the parent kernel down to obare minimum, reducing the chance of errors.

eg: you want to run an experimental iptable module on one of the virtual servers, no problem, if it crashes, all the other servers keep on trucking.

Essentially Xen provides a better sandbox from a stability/security perspective.

Re:Why Xen and not vservers?

[NitsujTPU]

Xen is 100% different. Also, Xen supports over 100 VMs per machine.

Also, Xen does things to make that 1 copy of glibc a reality. Arguably, that 1 scheduler is one of the primary reasons you would prefer Xen.

Re:Why Xen and not vservers?

[Stephen Williams]

Xen can be used to run different kernels side by side. So you could run, say, Linux and NetBSD concurrently on the same box.

-Stephen

Re:Why Xen and not vservers?

[tbcpp]

We have a X terminal server here at work that uses vservers. Doing it over again, we would probably use Xen as opposed to vservers. The vserves have weird bugs once and awhile. Such as processes not being able to talk to eachother for nor apearnent reason. The entire system will work fine for weeks, but once and a while there will be a problem that nothing short of a restart of the vserver will fix. If Xen works, and you aren't short of resources, use Xen by all means.

Re:Why Xen and not vservers?

[zeromemory]

While Xen appears as a neat package, why choose Xen instead of vservers?

Perhaps because vservers lack some of the neat features of Xen, such as on-the-fly instance migration and full iptables support?

Furthermore, vservers is, for the foreseeable future, a Linux-only project. So far, NetBSD and Solaris have been ported to Xen, and basic support for FreeBSD as a guest host is available. Once Intel VT and AMD Pacifica are available, Xen will also support Windows XP SP2.

Given just these benefits (and Xen has many more), it's no surprise that Xen appeals to more people and applications.

Re:Why Xen and not vservers?

[oddfox]

Aside from your post indeed being interesting and informative (It would be even moreso with more information related to exactly what's going on), I can't believe you'd choose a sig like your own. Do you have some sort of vendetta? Are you Fred Phelps or just have a sick obsession with what other people do in their private lives? Sheesh.

The Penny Arcade formula always proves true.

Re:Why Xen and not vservers?

[tbcpp]

The mantra of the "Christian" Homosexual movement is that the Bible does not condemn such acts. So, I'm basically using my sig as a way to express a different (and the correct) view on this issue, and stating that God finds such acts totally offensive. And now feel free to mod me down as being off topic.

Re:Why Xen and not vservers?

[Anonymous Coward]

Since you obviously have seen him, next time God talks to you, ask him whether he drives a BMW, Mercedes, or Lincoln. Most of the high-rolling evangelists seem to prefer Lincoln's or Caddy's, I'm just wondering if thats because God told them thats what he drives?

Re:Why Xen and not vservers?

[mikefe]

[...] Arguably, that 1 scheduler is one of the primary reasons you would prefer Xen.

Since I haven't read up on the details of XEN, I must ask the question.

Does that mean the host OS is the one scheduler in question?

Re:Why Xen and not vservers?

[elhedran]

With Xen, a kernel panic effects only that kernel, the other kernels keep on running. Under vservers, it takes down the machine.

Ok, why not User Mode Linux then. Its already in the kernel, and if an instance crashes its no worse than any other linux thread going down. And you can run linux binaries unmodified in it now.

As far as I can tell the main thing Xen gives is it can emulate more than just Linux, and has easier network setup. Another advantage includes easier migration of a Xen instance to another host. However I prefer UML since I only run virtualized Linux boxes and I'm more concerned about emulating hardware rather than just passing hardware out to the host machine. That and its currently in the kernel, and currently doesn't require a special toolchain to avoid special instructions.

Re:Why Xen and not vservers?

[Sloppy]

That's fine if the guest OS is Linux. But what if I want to test software on OpenBSD without rebooting?

Nothing against vserver or uml -- they're cool -- but they just aren't enough for some things.

Re:Why Xen and not vservers?

[NitsujTPU]

Sorry. I wasn't terribly clear.

Full virtualization in a virtual machine monitor can allow for the absence of a host OS. There might be an OS from which you tweak a few settings that acts as a "host," but it isn't a "host" in the sense that people think of in VMWare. VMWare has a product called ESX that works much more like this role.

Instead, what you would have is a virtual machine monitor that manages context and virtualization of the devices. It might merely multiplex devices for which this is an option. It might borrow drivers and offer a virtualized version. In this case, each OS has its own context (copy on write and other optimizations notwithstanding). So, each OS has its own scheduler, unlike in vservers, which share a scheduler.

Optimizations allow for this to be done nearly in the same light as merely flipping between applications or other processes, but since each OS gets its own scheduler, a high priority job in 1 OS, or forkbomb, or anything else, only affects that one virtual machine.

Xen has been used to do some very cool things that allow for the high scalability. Interestingly, however, it's a bit of a battle between the Xen guys an the VMWare guys, so there's enough propaganda and mudslinging in the process that you really need to sort it all out in your head to make an objective analysis. The first paper that I read on Xen touted paravirtualization. The last paper that I read on Xen was run in what sounded like a fully virtualized VMM.

Re:Why Xen and not vservers?

[sjames]

While Xen appears as a neat package, why choose Xen instead of vservers?

The right tool for the right job. With Xen, you can migrate a VM to another physical machine. Even a kernel exploit will only trash one virtual machine. Each virtual machine can have a different kernel if desired.

If your application doesn't call for that, vservers may be the right tool for you.

Re:Why Xen and not vservers?

[shani]

Ok, why not User Mode Linux then.

So I guess for some reason you just don't like Xen.

Anyway, to answer your question, UML is dog slow. It's not so bad for basic processing, but anything that requires disk I/O is basically unusable.

Irony?

[Koos Baster]

WTF?

I though irony was like rain on your wedding day, like a free ride that you've already paid, like the good advice that you just didn't take. Who would've thought...it figures?

Re:Irony?

[ettlz]

In my experience, there's very little difference between Jagged Little Pill and an XP install disc. The lyrics of "You Oughta Know" seem all too relevant.

Re:Irony?

[Anonymous Coward]

Thats not irony, its a simile...

Re:Irony?

[pecanNZ]

BLACKADDER: Baldrick, do you know what irony is?
BALDRICK: Yeah, it's like tinny or brassy, only it's made of iron.

Re:Irony?

[Anonymous Coward]

Irony === diff(expected,actual) != NULL

Re:Irony?

[Unique2]

Well actually -- I'll let commedian Ed Byrne explain this one to you:

"I love that bit of musical parody. I'd love to see them do like a musical parody of ironic by Alanis Morrisette, they could do a really wacky version with some irony in it.That would be an odd notion wouldn't it?
Cause that song really gets to me for that very reason cause she wrote a song about irony and filled them with things that were supposed to be ironic and none of them were. they were all just unfortunate. I always think that song should of been called 'Unfortunate'. The only ironic thing about that song is that it's called ironic and it's written by someone who doesn't know what irony is. Fairly ironic when you think about it.

I'm going to coin a new word which is 'alanic', that's things that aren't ironic but you might think they were if you were a dozy canadian bint

I'm not being harsh, if you actually listen to the lyrics to the song. "Like a traffic jam when your already late"- that's not ironic it's just a pain in the hole that's what that is. When was the last time you were late for something, got stuck in a traffic jam and said "Look on the irony on this , there's irony for ya. I'll tell ya I was in a fierce ironic traffic jam the other day iIll tell ya. The irony was ninety."

No, there's nothing ironic about being stuck in a traffic jam when your late for something. Unless your a town planner. If you were a town planner and you were on your way to a seminar of town planners at which you were giving a talk on how you solved the problem of traffic congestion in your area, couldn't get to it because you were stuck in a traffic jam, that'd be well ironic, I'm sorry I'm late you'll never guess.

"It's like rain on your wedding day", only if marrying a weatherman and he set the date. I could go on and I will.

A no-smoking sign on your ciggerette break, that's inconsiderate office management. A no-smoking sign in a cigerette factory- irony. It's not a difficult concept Alanis. It's very rare you see a ironic no-smoking sign although if you ever see one of those that say thank-you for not smoking and you are. Fairly ironic

The best line in that song has got to be the line "It's like 10 000 spoons when all you need is a knife." That's not ironic that's just bloody stupid. How big is your sink alanis? We haven't got 10 000 spoons beween us have we? What do you need this knife for? To stab the bloke who keeps leaving spoons all over your house. But we'll give her the benefit of the doubt. Imagine you needed a knife for something, couldn't find one cause all you find was 10 000 spoons, could happen. And therefore you couldn't do whatever it was you needed the knife for then the next day it turned out that a spoon would have done.

Re:Irony?

[tabrnaker]

rant rant rant rant. We all like to criticize the clueless. But then again, how much money are these people making?

It's not a thesis, it's a song, and guess what? It did exactly what it was meant to do which is sell and make money.

I wish we were paid in accordance with our knowledge. To some extent we seem to be, too bad it's usually an inverse relationship.

NetBSD

[Anonymous Coward]

Well, maybe Red Hat is pushing it to be first. On the other hand, Xen is pretty portable and NetBSD is by far the fastest host OS for Xen.
NetBSD Xen is a platofr officially supported by NetBSD, I think v3.0 will include it in the generic kernel even.

Mandriva 2006 includes Xen

[general_boy]

Mandriva Linux 2006 includes xen0 and xenU-enabled kernels and the Xen supervisor utilities package. The Community version of Mandriva 2006 can be downloaded from many Linux mirror sites.

I'm running such a box now with a total of three Linux domains (one host domain and two guest)... much easier than manually patching everything.

Re:Mandriva 2006 includes Xen

[N7DR]

Mandriva Linux 2006 includes xen0 and xenU-enabled kernels and the Xen supervisor utilities package. The Community version of Mandriva 2006 can be downloaded from many Linux mirror sites.

But unfortunately it's broken on 64-bit machines. (see bug #18432 at qa.mandriva.com; Mandriva says that they will upload the x86_64 glibc-xen packages when they have verified that xen works with their x86_64 kernels, which they currently don't).

Re:Mandriva 2006 includes Xen

[beef3k]

So does Fedora Core 4:

kernel-xen0-2.6.13-1.1532_FC4.i686.rpm
kernel-xenU-2.6.13-1.1532_FC4.i686.rpm
xen-2-20050823.i386.rpm

The whole point here is to push Xen into the mainline kernel, not just to provide it with the distribution as an alternative kernel build.

ITLB/DTLB

[Anonymous Coward]

How about fixing Xen to properly emulate the Intel x86 (post-Pentium-Pro) split-TLB architecture, so things like PaX will work in the VM? QEMU and Bochs also fail to do this, opting to use a single TLB instead. Until this is implemented, I'll be forced to continue using VMWare.

Re:ITLB/DTLB

[SirTalon42]

RedHat probably doesn't see that as a priority since they use SELinux in their kernels (which I guess doesn't use that).

All hail the new devil

[dascandy]

Virtualization technology is a very good thing. It allows you to use multiple operating systems at once, without fights for hardware control (which is why VMWare doesn't do it like this). But, if it's doable in hardware, it's doable in OS level software. Why didn't anybody do it then?

Put differently, how are AMD and Intel going to make it work? Since hardware doesn't like multiple masters (try a PS2 mouse with 4-5 byte protocol, it completely freaks out with a KVM switch), it's going to go haywire if you have multiple masters. Unless, of course, you don't. If all is virtual, make the entire computer virtual and let the OS meddle in the virtual space that's left after that. Include drivers for anything you like in the virtualizer core and let that start up the "OS".

The OS can then not use all features of the hardware, but only those given out by the virtualizer core. There's no escaping it, since it's the replacement for your BIOS. Combine that with the concept of Trusted (Treacherous) Computing and mix in a bit of Fritz chip and BIOS replacement Intel is pushing (as well), what do you get?

A computer with only drivers and everything (yes, EVERYTHING) in a checkable state, in which your OS doesn't even control the computer anymore, but the virtualizer. You don't want to change that one, since it renders your computer unusable. You do want to change that, since it renders your computer unusable. The state in which you control anything on your computer is becoming a paradox.

*puts on tinfoil hat, goes back within cage of faraday in radio-silent zone*

Re:All hail the new devil

[antifoidulus]

The powerPC is fully virtualizable, thats really how mac on linux works. It's x86 that lacks virtualization support....

Re:All hail the new devil

[davidsyes]

So, what have we got here? "Virtual Hell"? or "Literal Hell"?

Hmmm, I suppose virtual/animated snowballs now CAN stand a chance in "hell"...

How does virtualization work?

[DoofusOfDeath]

It basically lets you run multiple instances of the OS concurrently, where each instance thinks it's the only one running on the computer, right?

But then what do you when two or more OS instances want to monkey around with hardware that has state? For example, if one OS wants the screen resolution to be 640x480, and another OS wants the screen resolution to be 1024x768, you can't very well keep switching the screen between those two resolutions every time you change which OS is getting CPU time. Or another example is with printing: you can't very well interleave the print data streams from two OS's to the printer without hosing the print jobs.

Re:How does virtualization work?

[LiquidCoooled]

Print jobs are not an issue, the printer is just busy and the OS in question queues it.
As for varying screen resolutions, it just made me think of the Amiga and pulling down a low resolution screen to reveal a high res on behind.

very very cool stuff, but inherantly impractical, I imagine it will most likely be similar to the KVM switching or simple desktop switching?

Re:How does virtualization work?

[peterpi]

That's not a problem with linux; you take what X decides is good for you and thank your lucky stars if you don't have to edit your config file ;)

Re:How does virtualization work?

[digitalhermit]

Virtual machines also have virtual screens that are independent of each other. You can, for example, have a 800x600 window right next to a 1024x768 window. Depending on how you have it configured, toggling between full screen sessions of the VM will either re-size the screen or play inside a portion of the existing screen. It's still virtual video, however, so there's no conflict.

For printers you can either set up a print server or the printer gets attached to a particular OS instance.

Re:How does virtualization work?

[Waffle Iron]

But then what do you when two or more OS instances want to monkey around with hardware that has state?

The Schrodinger Corp. makes special PC cases that can handle those requirements.

Re:How does virtualization work?

[surprise_audit]

IIRC, screen resolution issues are handled in Xen by only letting the host OS set it on the display. The various guest OS's are accessed via VNC within that. Check out the Xen demo CD at: http://www.cl.cam.ac.uk/Research/SRG/netos/xen/dow nloads.html. You get a Debian system with some friends:

The Xen demo CD is a live ISO CD running Debian Linux that enables you to try Xen on your system without installing it to the hard disk. It enables you to start other guests running Linux 2.4 and 2.6, NetBSD and FreeBSD. Xvnc is used to enable the graphical console of the domains to be viewed.

Obviously it's not particularly speedy when loading binaries off the CD, but it gives you an idea of the potential.

Re:How does virtualization work?

[Anthony Liguori]

It basically lets you run multiple instances of the OS concurrently, where each instance thinks it's the only one running on the computer, right?

Sort of.

There is really two parts to virtualization: 1) processor virtualization and 2) hardware virtualization. Processor virtualization allows multiple virtual machines to think they have the whole processor. In this case, there's a lot of interesting tricks you can do (and new hardware support) so that the virtual machine thinks it has the actual underlying processor.

Hardware virtualization is really a relatively immature area. In general, you cannot virtualize hardware so you almost always end up emulating hardware. This is what let's you create a Window that the Virtual Machine thinks is it's VGA card. The VGA card is entirely emulated in software so you can make it do pretty much whatever you want.

Re:How does virtualization work?

[Doppleganger]

He asked about when they wanted to monkey around, he didn't say anything about cats.

Re:How does virtualization work?

[Just Some Guy]

Or another example is with printing: you can't very well interleave the print data streams from two OS's to the printer without hosing the print jobs.

A common FreeBSD setup is to run a print server in the host OS and configure jails (FreeBSD's virtualization systems, which are completely unlike Xen) to speak to that server. In short, you treat the jail environments like standalone machines on a network. I'd suspect you'd do something similar under Linux.

Re:How does virtualization work?

[Anonymous Coward]

It basically lets you run multiple instances of the OS concurrently, where each instance thinks it's the only one running on the computer, right?

But then what do you when two or more OS instances want to monkey around with hardware that has state? For example, if one OS wants the screen resolution to be 640x480, and another OS wants the screen resolution to be 1024x768, you can't very well keep switching the screen between those two resolutions every time you change which OS is getting CPU time. Or another example is with printing: you can't very well interleave the print data streams from two OS's to the printer without hosing the print jobs.

Poke your head out of the box for a second and think - how do you run multiple x sessions with different bit depths and resolutions, oh and dont forget users? Its not like your monitor is switching between the two every time one gets a few ticks. As for printing, havent you heard of spooling? Its this cool new thing. Everyone can print at once and the printer or the computer attached to it saves it in a queue until it can print. Have you never printed a huge OOo doc and then tried to print a fandango ticket? ZOMG - they both print! Oh and you didnt mention connectivity. How do you send bits down the wire at the same time? Think veth0(the v is for virtual). Everything your bitching about is handled by the Xen0 OS. The XenU OSs, or clients, contend for cpu/print/network time just like programs do now on a normal box.

Re:How does virtualization work?

[NereusRen]

"The Schrodinger Corp. makes special PC cases that can handle those requirements." (Score:3, Insightful)

Methinks the mods aren't paying too close attention today... Maybe they are both asleep and awake at the same time?

Re:How does virtualization work?

[dbIII]

if one OS wants the screen resolution to be 640x480, and another OS wants the screen resolution to be 1024x768,

Silly old buffers - you can do all of that stuff in memory. The host kernel handles scheduling, so makes sure that data gets to the hardware in a useful manner.

The video resolution thing isn't a big deal anyway. I'm currently running a conventional version of linux with three active displays but only one monitor. One is 16 bit colour, one 8 bit colour and the other is a fake X windows display exported to other machines via VNC. I can easily switch between two of them and display the third in a window, but as far as applications know they are real displays that are there all of the time. It's the same in a virtual environment,the host OS is in charge of doing things like putting the stuff from the OS the user wants to see on the screen, while the other ones don't get to touch the video card until you want them to.

Re:How does virtualization work?

[sjames]

For example, if one OS wants the screen resolution to be 640x480, and another OS wants the screen resolution to be 1024x768, you can't very well keep switching the screen between those two resolutions every time you change which OS is getting CPU time.

TGhere are several approaches to that, all of which virtualize and/or emulate the hardware. In your example, X might be set at 1024x768 witgh the othger OS displaying as a window. It could also be done similarly to Linux VCs where some escape command will switch which virtual VGA is actually displayed.

What?

[tomstdenis]

Multiple OSes at once? Isn't this a step backwards for the push for portable software?

I never got the whole concept of virtual servers..... this is why they invented vhosts and servers which could bind to multiple IPs ...

Tom

Re:What?

[Dr. Evil]

It's easier for sysadmins to say "sure, here's root, blow it up", than "no, you can't have root on this box, we'll consider your request for the patch during the next maintinence window"

It also lets you spawn dedicated servers in software.

For my home machine, my "server" can be stripped down to the most bare of bare components while my "sandbox" can have the compiler, web browser and other junk.

Re:What?

[LinuxHam]

Multiple OSes at once? [..] I never got the whole concept of virtual servers

Its mainly an enterprise play.

If you're an old-timer UNIX admin, you may have difficulty understanding the point of server virtualization (i.e. multiple OS instances). In UNIXland, it has been normal and customary for several completely unrelated applications to run under the same OS instance, together servicing thousands of users. That never worked well in Windowsland. That being said, it didn't stop manufacturers from making staggering improvements in performace and capabilities of Intel servers. Companies grew to expect single Intel boxes to perform at the levels of large UNIX servers. The only way to achieve that in recent years has been to use industrial strength virtualization technology (basically, ESX). The boxes are beefier than ever, and ESX isn't cheap, but it works wonders.

For $50k, you can run 50 VMware guests on one very beefy box (not counting SAN), but you'd want a second for failover. For $75k, you can run about 100 guests on 14 blades in 7U (again, not counting SAN) and have the guests automatically migrate to the blades most able to run their workload at that moment in time. Ask a blade to come down for maintenance, and all the guests scatter to other blades before the blade powers off. Replaced a dead blade with a blank? Your systems management policies detect the new blank and automatically install ESX on it so guests can migrate back and evenly spread out the load.

Sounds crazy, I know, but that's a taste of what we're doing in the enteprise space these days.

Re:What?

[tomstdenis]

Your "desktop" shouldn't be your server if you really use the desktop [e.g. crashes because the latest nvidia driver isn't stable isn't a good thing].

In my case I do nightly backups of my home dirs and what not. In the grand scheme of things if my desktop got fucked up it would take me a days worth of building to get it back, If I'm lucky and I've done a recent ghost it can be back up in minutes...

I can see the whole VMware shit for windows servers but UNIXly servers like apache are meant to host multiple sites, they run on an OS that has multiple users, etc, etc.

Tom

Re:What?

[tomstdenis]

You don't need virtualization for task migration. OpenMOSIX is an example of that. It can send processes to any number of hosts and can detect when they're dead.

The VM game is because "everyone wants to be root" ...

In the same box where you host 100 users in VMs you can probably host 200 natively.

Tom

Re:What?

[Joe Fiction]

I agree, let them all run on a single OS... at least at first. The problem is application are quite happy to misbehave and bring down the other dozen apps along with it. If ever there where a phrase that pays, to apply here, it's "good fences make good neighbours".

Re:What?

[tomstdenis]

Well I can see it making sense if you are hosting non-friendly users [e.g. different companies].

If you're one company and you're hosting say 10 domains ... well you're not gonna try and cheat yourself?

As for apps misbehaving ... the flipside is you have to administer patches, updates, config fixes multiple times.

Tom

Re:What?

[LinuxHam]

You're right about the grid-like technologies, but I wasn't referring to that. In the enterprise space, VM stuff is to drive up utilization. You hear the "root" story from developers. Large enterprises do it for cost. Its actually less expensive to run servers as guests and you get high availability as a throw-in.

Re:What?

[Dr. Evil]

For a home machine, crashes are just fine. Remote exploits are not.

I run my home server headless and use Cygwin/X for remote apps anyways. It gives me a few Linux apps on my Windows machine and keeps away the kind of instability you describe. IMHO, Windows is by far a better desktop OS than Linux... but Linux is a better server and has some fantastic tools.

Re:What?

[tomstdenis]

two words. "multiple desktops".

Oh yeah and "natively".

I can't stand to sit at a windows desktop. It infuriates me. The "cmd" shell is inept, the desktop is one-paned, etc...

It may do fine for non-developers and/or people with money. But X11 works much better for developres and people who spend their money elsewhere.

Tom

Re:What?

[Dr. Evil]

The multi-paned desktop environment seems to be required only because X11 doesn't have a concept of maximized windows or decent keyboard shortcuts to move between apps non-visually.

A terminal shell to a Linux machine is just as good as a local shell on a Linux machine. Linux is awesome at network transparency... Using a Windows client and a Linux server plays to Linux's strengths and Windows' weaknesses. You get the best of both worlds.

You seem to be suggesting that a dedicated Linux server and a dedicated Linux client is better than a dedicated Linux server with virtualized Linux sandboxes and a dedicated Windows client. I can't agree. Your argument seems to center around the notion that Windows is useless though. I can't agree with that either.

Re:What?

[fimbulvetr]

I'm not planning on using xen for this, but think of the advantages on say, 12 identical mail servers.
All I have to do is patch the kernel once and reboot the servers individually (since they're load balanced). What's more - I can have a virtual server with the newest postfix snapshot that's otherwise identical to the debian on my servers. I can test it to my hearts content and decide whether I want to go production with it or not. All without much hassle at all - it's not like I have to build another machine/provision another switch port/rack another 1u/etc.

And, not only that, but I could have like 3 instances on 4 different machines. I could easily move any of the three instances around on the machine, and move any of the three to to any of the other 4 machines! Now imagine scripting this so - say - you could be in the bar at 1:00AM and not have to worry about "that call".

Now, picture using this for say, oracle, apache, etc. It keeps sessions! Right now, if one of my foundry load balanced apache servers fails, the sticky ssl sessions are going to lose their session - but not with virtual servers - it can move them to another physical server with very little latency.

Check out what secureboot was saying:

http://linux.slashdot.org/comments.pl?sid=166960&c id=13922864

And check out the second (pdf) link in this post:
http://linux.slashdot.org/comments.pl?sid=166960&c id=13925022

Imagine

[mrjb]

Imagine if you would the ability to use Xen for unlimited operating systems
Are you asking us to imagine what I think you're asking us to imagine?

Re:Imagine

['nother poster]

No, he's not. I swear to you. Now, put down the cliche, and back away slowly.

XEN vs UML

[vlm]

Could someone clearly explain the difference between XEN and user mode linux beyond "red hat supports xen"?
I've been a happy paying customer of a UML provider (linode.com) for like multiple years.
The description of XEN "sounds like" a description of UML to me.
So is XEN just a reinvention of UML or what?

Re:XEN vs UML

[secureboot]

One big difference (from a user perspective) is performance. Check out the xen paper here . Xen performance is often 4x better (literally) than UML because it paravirtualizes, instead of fully virtualizes. Sorry I don't have time to get into technical detail, but that's one quick answer.

Re:XEN vs UML

[Feyr]

UML: now with 30% more performance

Re:XEN vs UML

[Octorian]

"User-Mode Linux" is a Linux kernel patched to run as a user-mode application. It is not a virtual machine.

Usermode Linux already in the kernel.

[nietsch]

These guys(Xen) have all these companies donating money to them, but have been beaten to kernel inclusion by UML. UML is basically a two man project, developed by Jeff Dike and Paolo Giarrusso (aka Blaisorblade). Xen may be multi platform and all, but thus far UML is easier to handle and does not require the host to run a patched kernel (you could use a patched kernel, but the newest development Skas0 does not need it).

Re:Usermode Linux already in the kernel.

[Anonymous Coward]

That's the whole point. XEN has the potential (once AMD/Intel have their CPUs updated) to run unmodified OSs. In other words, your main linux box will be able to host any x86 OS within XEN, each one independent of each other. UML, on the other hand, is a linux kernel binary which executes in user-land, thus you only have linux within linux.... mind you, uml is great and i've used it in production mode capacity..

Note, not all solutions fit the same problem, nor are many problems solved by one solution. Just pick the one you are in need of, and run with it. This is a good thing.... it's about having more choices.

Re:Usermode Linux already in the kernel.

[zeromemory]

These guys(Xen) have all these companies donating money to them, but have been beaten to kernel inclusion by UML.

Being the first to the party doesn't always mean you're going to the best; see DevFS vs. udev.

Xen has much greater performance than UML and supports more operating systems. While UML is currently more mature and stable than Xen, it's only a matter of time before Xen surpasses UML as the preferred virtual server technology. Hell, even Linode, a strong proponent of UML technology and virtual server hosting provider is migrating to Xen.

FYI, I'm currently running a Xen-based system with 15 virtual server instances for a system administration course at UC Berkeley on a server built with cheap off the shelf components (AMD Athlon 64 2800+, 1 GB RAM) and everything is quite snappy. It'd be difficult to even approach such usability with UML, and I'm using Xen 2.0.7. I can't see what Xen 3.0 will bring.

Re:Usermode Linux already in the kernel.

[iabervon]

They apply to somewhat different situations. UML is more for the case where you want to run a virtual environment under a normal environment, while Xen is more for when you want to run a bunch of parallel virtual environments. There's a certain amount of overlap in usage, but there are enough differences that both are worth having in the kernel, like, for instance, the loopback device and device mapper (both for the situation where you want a block device that isn't exactly a piece of hardware, but you don't build disk images with dm or do RAID on loopback). Ideally, UML and Xen will be able to share code for the common stuff (e.g., the virtual hardware).

Re:Usermode Linux already in the kernel.

[OverlordQ]

Get back to me when Xen supports PAE. kthx.

Re:Usermode Linux already in the kernel.

[mikefe]

PAE support is already in the 3.0 codebase that is pending release.

If you are interested, it's probably already in the various distro development trees (red hat rawhide, mandr(ake|iva) cooker, Novell/Suse ... (what do they name their ongoing dev repo?)).

Try your most convoluted setup now, and know it will be ready for production at release time.

Re:Usermode Linux already in the kernel.

[hendersj]

Novell/Suse ... (what do they name their ongoing dev repo?)

Do you mean OpenSUSE?

Re:Usermode Linux already in the kernel.

[mikefe]

You notice how I mentioned Red Hat/Rawhide?

Rawhide is the development repository where every new package goes when it is first created. Just like Mandriva/Cooker and Debian/Sid.

Now that my question is further defined, can someone answer it?

Re:Usermode Linux already in the kernel.

[hendersj]

Try the YaST Package Repository on ftp.suse.com. On the OpenSUSE site there are links to mirrors of the development build, which is where I imagine you'd find all the latest stuff being tested in the current alpha/beta builds.

Re:Usermode Linux already in the kernel.

[mikefe]

Does that tell me the code name for the Suse development repository?

Re:Usermode Linux already in the kernel.

[hendersj]

No, it tells you where the repository is. Not everything has to have a "cool" code name to be useful.

Re:Always FOLLOWING Microsoft ! Think NEW !

[Anonymous Coward]

"And which big iron hardware has been doing since the 1960`s"

Would it be the same big iron that is after SCO?

Microsoft leading?

[Tominva1045]

This move comes as Microsoft is pushing its own virtualization products and recently relaxed some of its licensing requirements around Windows Server 2003 to facilitate more pervasive adoption and use of those technologies.

Did it take Microsoft actually doing it to make this a priority?

Should we be thanking Mr. Bill?

Re:Microsoft leading?

[Anonymous Coward]

Parent is borderline insightful (a little on the obvious side to be truly insightful, sorry). And on Slashdot it's +5 insightful.

The competition between Microsoft and the open source crowd is really heating up and we've been getting a lot of movement and some exciting things coming out of both camps. It's really a cool time to be into IT and CS right now (unless you care about money).

Re:Microsoft leading?

[The_Dixie_Flatline]

Noooooooooooooo?

MS leapfrogging

[MobyDisk]

I know everyone complains about how MS lacks innovation, but this is a good example of BUSINESS innovation. Virtualization isn't new. I've used it before, seen it before. But MS bought an existing product, then wrapped it up nice and pretty and easy, and presented it as a solution to a major problem. And it is getting widely adopted. My office uses virtual servers constantly to simulate production environments for development: it saves time, money, and effort.

I never even considered virtualization of servers or development environments until I learned about MS Virtual PC and MS Virtual Server. Norton Ghost or dd dumps were all that I knew. So Microsoft is doing something right, and they will be perceived as the innovator and the winner here. They will be selling that you can virtualize servers to save time and money, and companies will buy it. They won't even know that this originated in the *nix world.

I look forward to seeing what the next leap in this technology is. I suspect we are just beginning to see some novel uses for it.

Re:MS leapfrogging

[sw155kn1f3]

Nevermind that my company used vmware for exactly the same purpose you describe here for 6 years already.

Re:MS leapfrogging

[dodobh]

They won't even know that this originated in the *nix world.

It didn't. This is mainframe technology. It just didn't work very well on x86 and hence the Windows using world was unaware of it. There is quite a bit of stuff that doesn't exist in the Windows world yet.

Re:MS leapfrogging

[Anonymous Coward]

yeah, the old IBM 370's running VM were great at this... you could spin up an entirely seperate virtual machine, and run your test kernel with custom assembly code drivers, etc, and the worst you'd do is crash your virtual machine. This concept has been around since at least the 70's, its just new to the commodity PC world.

Re:MS leapfrogging

[TheRaven64]

I went to a couple of talks by IBM on Xen and the virtualisation tech in the Power5 family (which, by the way, is very shiny). My favourite quote was:

`We had an advantage over other people doing virtualisation in that when we got stuck we could pop along the hall to the mainframe guys, and say "you remember 20 years ago when you had this problem? How did you solve it?"'

SuSE has had Xen since V. 9.3

[cypherz]

http://www.novell.com/products/linuxprofessional/c omparative.html

Re:SuSE has had Xen since V. 9.3

[zeromemory]

Yeah, but having Xen in the kernel mainline gives the project much more credibility and exposure.

A problem with Xen has been facing is keeping up with all the changes occuring in 2.6. If Xen is merged into mainline, there's a much better chance that Xen will be able to support the features and bug/security fixes that get added to 2.6 with each release.

For example, the current Xen stable (2.0.7) supports kernel 2.6.11.12. Every time a new security hole is discovered, system administrators using Xen have to manually backport a fix from the latest kernel. Having Xen in mainline should make this process much easier.

Re:SuSE has had Xen since V. 9.3

[cypherz]

>Every time a new security hole is discovered, system administrators using Xen have to manually backport a fix from the latest kernel.

Yeah, its problems like that with Xen that keep our shop on VMware.

It's not personal, it's just marketing

[FishandChips]

The Slashdot summary is a bit misleading. What the article says is that Andrew Morton has been expecting a kernel submission for Xen for quite some time now but a) has yet to receive it, and b) needs to go through the usual process with other "stakeholders" before any incorporation. Later the article quotes the Xen folks themselves who point out that "feature creep" and the need to generally get things really solid and stable has made everything take a little longer.

What the article actually seems to be saying - it uses the word "agressive" a lot as if this was some kind of virtue - is that Red Hat has a new senior honcho who'd like to make his mark. The issue of incorporating virtualization technologies into the Linux kernel is taken as a given by all parties. Which is hardly news. Chalk one up to the Red Hat marketing department for a nicely planted "news" story about their increased investment in the area (new hires, etc.), perhaps.

Re:It's not personal, it's just marketing

[Anthony Liguori]

The Slashdot summary is a bit misleading.

As is the article. Let me try to clarify:

Andrew has been open to including Xen (after appropriate changes) in his tree for quite a few months. The Xen developers have been busy working on the 3.0 release (which has finally tapered off but is a very ambitious release). Just a couple of weeks ago, the Xen 3.0 Linux port was forked into a new tree so that there could be a focused effort on getting it into shape for inclusion by Andrew.

The currently plan is to try and aim for 2.6.15. That's most likely the earliest it could go in. One of the difficulities in preparing the Xen port for inclusion in the kernel is that it touches a lot of Linux and requires a ton of kernel knowledge and a bunch of people to agree that it's the right thing to do. Getting key kernel guys involved (like the kind at RedHat) is critical to have this happen.

"Innovated" by wb:

[wild_berry]

I expect that any self-respecting adware/spyware/rootkit maker will hide the bulk of their work out of sight in a virtualized environment. Were I designing a zombie clone, it would hide out-of-sight ona virtualised machine -- a no-brainer because it's harder to discover. And then people can have fresh installs of Windows which are patched up still resulting in their 14-month old 4.2GHz Pentium crawling and needing replacing.

Re:Two birds, one stone... VMware

[Anonymous Coward]

>>> get some market share from VMWare.

(FTFA: he [Morton Andrew, 2.6 kernel maintainer] said. "VMWare is also working on virtualization in general, and they will provide feedback on the proposed design.)

  Sounds very odd to me that Andrew needs feedback from VMware. You would think that VMware would provide the following feedback: "SoaB!?!?!"
  Why would he do VMware any favors? This is in competition with VMware and VMware are big boys who can handle thier own business model.
  You, like myself, should be a little confused about this.

Xen and UML?

[MrWiggum]

What does everyone think? If the kernel developers decide to include xen in the kernel will they drop support for UML? I think they should decide which is more used and include only that one to keep the kernal from becoming bloated.

Re:Xen and UML?

[some_random_person]

That's a good idea. While we're at it, let's get rid of all those unnecessary Linux distributions, too.

The kernel only begins to get bloated when you decide to compile in support for features/hardware that you don't use or have no intention of using.

Of course

[RandomPrecision]

In desiring to put Xen in the kernel, they have already failed.

Re:Of course

[Anonymous Coward]

What is the sound of one virtual OS kernel panicing?

Xen Virtualization HW "acceleration"

[emj]

Both AMDs Pacifica, and Intels "VT" (Virtulization Technology, used to be called Vanderpool) is getting support in the up comming Xen 3.0 release in december (hold your thumbs). So we will perhaps see some serious boost up in performance, there was recent discussion on the mailinglist about it. I'm waiting for Pacifica support which seems to be abit better (DMA virtualization), but that willl probably not be in XEN 3.0.

• AMD technotes about pacifica virtualixation
  
• AMD64 Architecture Tech Docs (pacifica PDF on top)
  

Re:Xen Virtualization HW "acceleration"

[Chirs]

The article specifically states that support for Pacifica will not be in Xen 3.0.

Re:Xen Virtualization HW "acceleration"

[tupshin]

One of the main Xen developers has stated on the Xen devel mailing list that Pacfica support will probably be in 3.1, which should still come out before the general availability of Pacifica enabled CPUs from AMD around the middle of next year.

-Tupshin

A solution to patches breaking functionality?

[div_2n]

Could this type of virtualization be the stop gap solution to security patches and updates breaking things? So a new patch is released for something? Clone your working environment into a virtualized one and apply the patch to that. Work under the patched environment and if after X amount of time the patch shows no signs of breaking things, commit the patch to the base install.

Seems like a good idea to me unless I am missing something.

Re:Two birds, one stone... VMware

[bernywork]

Keep your friends close, and your enemies closer?
Here is a garden path, care for a walk?

You never know, VMWare might like the idea of booting MS out of the market all together and keep Xen on the small end of the playing field and build tools to migrate Xen machines to VMWare machines. (I don't know if this exists already)

I remember that Xen although technically capable of emulating a base platform for Windows isn't allowed to now because of patents. With the new virtualization in the chip, I think the Xen developers are hoping to be able to run Windows over the top of Linux. It could be for this reason that VMWare is working with Andrew Morten and vice versa to make sure that Xen doesn't step on VMWare's toes.

Xen looks like a bloated pile of horseshit to me

[Anonymous Coward]

Seriously. I hated their webpage, their descriptions and the difficulty of getting the silly thing to run.

I'd rather see Linux VServer included

[gtrubetskoy]

There are a few problems with Xen. First, it's i386 only. Second (and this is the biggest problem IMO) - Xen is venture-backed, and seems to be extremely eager to show their investors a return. Nothing wrong with that, but it's important to consider the motivation, and the consequence of a funding pull back. If XenSource does not turn out to be a great business, then will Xen still be developed and maintained? Why not wait a little bit, in the open source world quality over quantity matters and time pressure should not influence development.

Also, there is another project that I plug every chance I get - Linux Vserver. Unlike Xen, this is a purely volunteer effort, and is very innovative and attemtps to solve a difficult issue. Unlike Xen, these guys actually do not want to be in the mainline for now, becuase they think it will slow down development. Because Linux VServer is taking a different approach to virtualization (better known as separation, which was pioneered by FreeBSD jails and is also now supported in Solaris), the end result is cross-platform, i.e. runs on any architecture that Linux runs on.

Now in the past whenever I posted about Linux VServer a lot of folks said that Xen allows you to run multiple operating systems and that that is why it is so useful. I think that in reality running multiple OS's isn't all that valuable - the only case where it may be very useful is software development, but that's a tiny fraction of the Linux users. We've been using Linux VServer for hosting, and we are absolutely convinced that this is the right solution - for using Xen for example would introduce all kinds of problems (starting with resource bloat).

Yet unfortunately the OSS world has become PR driven lately. Very few people are technically capable of looking at things based on its merits and just go after the things that have the most buzz, not realizing that the buzz is artificially generated.

Re:I'd rather see Linux VServer included

[adrenaline_junky]

I think the benefits of being able to run multiple operating systems are more numerous than just the single case of development that you listed. For a hosting company, the ability to provide multiple operating systems provides a great deal of flexibility. And for users, it gives one the abillity to check out various new distributions easily, as well as the potential to use different distros for specific tasks for which they are best suited (for example, perhaps you really like the VPN support of a particular distro).

Re:I'd rather see Linux VServer included

[tweek]

You have a misunderstanding of what Xen really is. Vserver is NOT the same thing. Vservers are nothing more than a fancy chroot and some other minor things. Mind you, I am NOT mocking or disparaging the project in anyway. There is a place for it but it does NOT create the same security domain that a Xen does. VServer is paravirtualization.

Xen isn't TRYING to replace the vserver market. It's trying to compete with the market of an IBM pSeries machine.

Until Xen came along, the only thing you had in the Intel market was VMWare ESX server or maybe the HP Superdome (not up to snuff on my Superdome)

Check these links for what Xen is trying to match:
http://en.wikipedia.org/wiki/Hypervisor
http://en.wikipedia.org/wiki/IBM_p5

Let me say this though, I agree that this is the outcome of large commercial interests in Linux (which I don't object to mind you). RedHat wants to beat Microsoft to market with a good virtualization product. Thus they are going to push hard to get this into the kernel so it's much easier to support (i.e. no backports or extra work involved beyond making it RHEL ready).

Re:I'd rather see Linux VServer included

[gtrubetskoy]

I think the benefits of being able to run multiple operating systems are more numerous than just the single case of development that you listed. For a hosting company, the ability to provide multiple operating systems provides a great deal of flexibility. And for users, it gives one the abillity to check out various new distributions easily, as well as the potential to use different distros for specific tasks for which they are best suited

All true. Let me rephrase the requirement here. Let's say I run an operation that requires a large number of servers. I know the application and I know the distribution I want. For example I am an ASP that hosts an application that does something and each instance is a separate server, let's say I need 10,000 of those. I want all servers to look exactly alike. I'm not going to play with distributions, I don't need separate operating systems, I am strictly after efficiency - I want to fit the most hosts onto a hardware box with minimal overhead.

I also want to have one filesystem to backup on the physical box, I want to be able to enter virtual servers seamlessly for management purposes. I want to have to maintain one kernel per hardware box.

In this scenario a separation solution such as Vserver, jails or Zones wins hands down.

Re:I'd rather see Linux VServer included

[gtrubetskoy]

Vservers are nothing more than a fancy chroot and some other minor things.

This is not a fair statement. I can say that Xen is not even as fancy as a CPU emulator. :-) Actually a "fancy chroot" is a very complex challenge - in fact the reason there are so many hardware emulators and not so many separation technologies is because implementing hardware emulation is simpler.

But yes, Xen and VServer are completely different beasts.

Re:I'd rather see Linux VServer included

[Anthony Liguori]

There are a few problems with Xen. First, it's i386 only.

Not true. Today, Xen supports i386, x86_64, and ia64. Xen is currently being ported to PowerPC also.

Second (and this is the biggest problem IMO) - Xen is venture-backed, and seems to be extremely eager to show their investors a return.

XenSource is a company backed by VC. Xen is developed by a much larger community though. There are a ton of press-releases that XenSource puts out that have the typical marketting junk that most Open Source folks despise but whatever, XenSource != Xen. Most of there people aren't even actively working on Xen anyway (they have a product for Xen management),

If XenSource does not turn out to be a great business, then will Xen still be developed and maintained?

Absolutely.

Also, there is another project that I plug every chance I get - Linux Vserver. Unlike Xen, this is a purely volunteer effort, and is very innovative and attemtps to solve a difficult issue. Unlike Xen, these guys actually do not want to be in the mainline for now, becuase they think it will slow down development.

Yup. That's why VServer is not in the kernel--they don't want to be in the kernel. VServer is a cool project, and I would love to see it end up in the kernel. Xen is also a cool project and it would be great to see it in the kernel. The kernel guys *will not* accept crap. Large portions of the Xen Linux port are currently being rewritten to live up to kernel standards. I have a ton of faith in the kernel folks overseeing the process.

Re:I'd rather see Linux VServer included

[Anonymous Coward]

Before Xen there was (and is) Usermode Linux. As I understand it, UML is slower (especially on I/O and other syscalls), and requires more guest modifications (you're not running the i386 instruction set, you're running the UM instruction set, that just happens to be almost identical. But you need a new kernel.)
Xen isn't quite as fully virtualized as UML, which is why it performs better. That may make it somewhat more vulnerable to problems, and it's much newer. YMMV. I run UML's in general for security and ease of maintenance, and on real dedicated hardware when performance is an issue. Xen falls between those two extremes. It'll either take from both, or be squeezed out.

Re:I'd rather see Linux VServer included

[asdfghjklqwertyuiop]

The main reason I don't like separation, at least as it was implemented with FreeBSD in my expereince, is the lack of complete resource separation. You can have one virtual "machine" go crazy and chew up all available memory, or use up all entries in the process table. Those are just a couple of examples. There are all kinds of other little shared, exhaustible resources that get shared between all VMs and permit one VM to interfere with another very easily. Like file descriptors, shared memory segment names for example. When all your VMs have their own kernel, none of that is a problem. Can vserver guarantee separation of *every* exhaustible resource in the kernel?

Re:I'd rather see Linux VServer included

[gtrubetskoy]

The main reason I don't like separation, at least as it was implemented with FreeBSD in my expereince, is the lack of complete resource separation. You can have one virtual "machine" go crazy and chew up all available memory, or use up all entries in the process table.

Yep, it looks like FreeBSD jail's development didn't get far past the basic implementation of jail id's. We started out with FreeBSD jails and then switched to Linux VServer because VServer was advancing while jails just kinda sat there. Where they differ is the VServer people look at each one of those limitations you described and rather than just saying "this is how it is, learn to live with it", they try to address it - so to prevent one context from chewing up CPU there is a token-bucket CPU scheduler, there are various memory limits, some interesting stuff being contemplated - e.g. using the rate of page swap-ins in combination with a token-bucket for fair memory access. There are lots of little issues, but that's what makes this project interesting.

Re:I'd rather see Linux VServer included

[mikefe]

Today, Xen supports i386, x86_64, and ia64. Xen is currently being ported to PowerPC also.

I would not be surprised if Sun ports it to Sparc also.

[...] XenSource != Xen. Most of there people aren't even actively working on Xen anyway (they have a product for Xen management),

Please read sig.

Large portions of the Xen Linux port are currently being rewritten to live up to kernel standards.

From what I have read, it is not that the code is low quality in XEN, it is that the XEN patch adds a new architecture which means a *lot* of code duplication. That has allowed the XEN team to try out many different ideas with fewer integration problems, but it is not acceptable for a merge into the mainline (kernel.org) code base.

Re:I'd rather see Linux VServer included

[Kashif Shaikh]

The Linux vserver sounds very similar to UserModeLinux in both capabilities and features. Both run linux with linux, but UML has been included within the kernel. Care to describe why vserver is better/different than UML?

Kashif

Re:I'd rather see Linux VServer included

[Anonymous Coward]

they are completely different - this article discribes the difference pretty well.

Re:I'd rather see Linux VServer included

[Anthony Liguori]

From what I have read, it is not that the code is low quality in XEN, it is that the XEN patch adds a new architecture which means a *lot* of code duplication.

You're right. The xen architecture is a problem. The drivers also need a lot of cleanup. It's not bad code, it's just not kernel code. The kernel guys tend to be fickle about how code looks.

valve

[uberjoe]

I don't know. I always thought Xen had too many jump puzzles to be included.

Re:valve

[Anonymous Coward]

I laughed.

Attention Red Hat. Step 1:

[Rogerborg]

Add "irrevocable and in perpetuity" to your 'defensive' patent 'promise'.

We don't want another SCO when you get bought out by Sun or Microsoft.

Old news

[Anonymous Coward]

Dupe. I heard this at my June TriLUG meeting (from a RH guy) and have since seen it on /.

Xen is the future...

[samj]

bring it on I say.

Re:Xen looks like a bloated pile of horseshit to m

[WhiplashII]

and the difficulty of getting the silly thing to run

Wow, if I were that dumb, I'd keep it to myself! (They have a live CD, you have trouble putting CDs in drives?)

Volunteer efforts can fall apart, too

[robla]

Why is it that you consider an all-volunteer effort inherently more robust? Key volunteers can have life changes (job change, health, etc) that cause their involvement to change. VC projects have the benefit of providing dedicated staff, professional project management, business development and marketing to keep momentum alive.

Yeah!

[Xenex]

It's been a long-time goal of mine to be integrated into a kernel.

Re:Two birds, one stone... VMware

[Wodin]

You never know, VMWare might like the idea of booting MS out of the market all together and keep Xen on the small end of the playing field and build tools to migrate Xen machines to VMWare machines. (I don't know if this exists already)

Well, VMware announced recently that they would support (or did in the latest version of VMware Workstation?) paravirtualised Linux (i.e. the Xen guest version of Linux.)

Re:Why Xen and not vservers? (OT)

[oddfox]

We are off-topic here but my observation had much more to do with how it's incredibly rude to even put that in a sig to propogate it wherever you post than why you put it there. This is a largely tech-oriented discussion site, quotes like that serve no purpose other than to draw attention that you most certainly have drawn.

I will say this, not only is that verse part of the hebrew scriptures, but it's translation is still in constant dispute. Nevermind the fact that these scriptures were written by humans, and the Bible has seen a large amount of edits over time, with exclusions and inclusions being made. It has been altered to suit the tastes of those in power in the religious structure.

As the anonymous coward questioned, when did you speak with God personally and discover his exact views on the subject matter, or do you just like to presume to know these things and spread bigotry without understanding the full context of the passage (Remember, context is always important)? Blindly following something is never wise, whether it regards religion or not, and this is not a slam on any form of organized religion. It's pretty strange to presume to know things one cannot possibly know, and claim that your preferred interpretation is the one true one. I'm not saying it's not true, but to proclaim it as such is foolhardy at best.

Besides, have we not all sinned, and isn't the ultimate purpose of the teachings of Christ to lead a good life? Cute how you put quotes around Christian as if you're any less a follower of the teachings because you interpret one thing differently, it really gives your contention that extra umph of truth and finality.

Re:Why Xen and not vservers? (OT)

[tbcpp]

I would normally agree with you here, however, I figure if some have the right to propagate their views of pornography or whatever they like, I should be able to express my views of "intolerance", as you put it.

And yes, I do talk to God on a regular basis, and he with me. And I do know his exact views on the subject of homosexuality. I agree that "Blindly following something is never wise" and that is the very reason why I have scripture, and personal experience to back up every view of morality that I have. The Scripture does instruct us after all to "have an answer to every one who asks of us a reason...". If you would care to discuss this at length we could begin a e-mail discussion.

Now that we are totally off-topic...

What's taking so long?

[Crazen]

Why isn't this here already?

Re:Two birds, one stone... VMware

[ckaminski]

Correction: It's not allowed to because of Copyright and licensing issues, not patents. The original WindowsXP XEN Guest was developed under an MS Educational Source License, and hence is not redistributable.

Re:Two birds, one stone... VMware

[bernywork]

I dug up the article about what mentioned Xen and patents. You are correct, that it requires a modified version of XP to run, but apparently patents are still an issue. So my guesswork still could be correct (Or completely wrong) in that VMWare may be involved so that the community doesn't tread on toes.