Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

We are sorry to see you leave - Beta is different and we value the time you took to try it out. Before you decide to go, please take a look at some value-adds for Beta and learn more about it. Thank you for reading Slashdot, and for making the site better!

The Book of Xen

samzenpus posted more than 4 years ago | from the read-all-about-it dept.

Programming 88

swsuehr writes "The Book of Xen: A Practical Guide for the System Administrator provides an excellent resource for learning about Xen virtualization. I frequently need to create test environments for examples that appear in various books and magazine articles (in the interest of full disclosure, I've never written for the publisher of this book). In the days before virtualization that meant finding and piecing together hardware. Like many readers, I've been using virtualization in one form or another for several years, including Xen. This book would've saved hours searching around the web looking for tidbits of information and sifting through what works and doesn't work in setting up Xen environments. The authors have done the sifting for me within the ~250 pages of the book. But far beyond, the authors also convey their experience with Xen using walkthroughs, tips, and recommendations for Xen in the real world." Read on for the rest of Steve's review.The Book of Xen is written with the system administrator in mind; someone who is comfortable with tasks like installing Linux and working with the command line. While it wouldn't be impossible for someone completely new to Linux to accomplish the tasks in the book, a bit of experience would go a long way to both visualize and complete the installation and configuration steps shown in the book. As stated in the introduction, the book is organized "(mostly) alternating between theoretical and practical discussion [because] an admin needs both practical experience and a firm theoretical ground to effectively solve problems..." (xxiii).

The authors do an excellent job of explaining what Xen is and where it fits in the virtualization landscape. This explanation begins with the introduction where the reader gathers a brief history of virtualization along with Xen's place in the landscape. Xen's limitations and reasons for using Xen are also covered right in the introduction, along with an overview of the book.

Chapter 1 begins with a high-level overview of Xen. This discussion is excellent if only to get the readers on equal footing for the discussions to come later in the book. Included in this chapter is a discussion of various techniques for virtualization including Full Virtualization, OS Virtualization, and Paravirtualization. The section on Paravirtualization leads nicely into some of the underlying details of scheduling, interrupts, and memory, and other resource management which are handled by Xen and discussed later in the chapter.

Chapter 2 sends the reader down the path of installing and using Xen. It's a short chapter, coming in at about 9 pages, and the reader is expected to be able to handle an install of CentOS with just a bit of guidance from the authors on specific options to select. This is a key point for those among us who have a preference for a certain Linux distribution. The book isn't tied specifically to a single distro, as the authors note in the introduction, "[w]e've tried to keep this book as distribution- and version-independent as possible, except in the tutorial sections, where we try to be extremely specific and detailed..." (xxiv). The base or host system upon which the examples run is based on CentOS, which the authors acknowledge and highlight in Chapter 2, "[f]or the purposes of this chapter, we'll assume you're installing CentOS 5.x with the server defaults and using its built-in Xen support. If you're using something else, this chapter will probably still be useful, but you might have to improvise a bit" (13). There is discussion of the Xen-tools package in a later chapter which shows its installation under Debian Linux too. So far from being tied to one distro, the book is refreshingly neutral in this regard.

By the end of Chapter 2, the reader has a working Xen host system and a domain 0 or dom0 host upon which to provision virtual machines. Included in Chapter 3 is a discussion of how to provision guest operating systems, known as domU in Xen-speak. The authors devote a good number of pages to making this task clear, and work through examples of basic domU installation and the use of package management systems and Debian's debootstrap to create domUs. Additionally in Chapter 3 the reader learns how to convert VMware disk images to a format usable by Xen.

Chapters 4 and 5 examine details of the Xen backend, including storage and networking. Chapter 4 stands out for its recommendation of blktap and LVM (Logical Volume Manager) as the storage backend as well as an overview of LVM itself, along with the use of networked storage for Xen.

Chapter 6 looks at tools for management of Xen, focusing on Xen-tools, libvirt, and Xen-shell while Chapter 7 gives advice for hosting untrusted users with Xen. Chapter 8 discusses the use of Xen with Unix-like operating systems and includes sections on Solaris and NetBSD.

The ability to migrate the virtual machine from one physical machine to another is one of the advantages of virtualization. As pointed out by the authors, a virtual machine might be migrated to take advantage of newer hardware, to perform maintenance, or any number of other reasons. Chapter 9 is of interest for its discussion of Xen migration. Cold and Live migrations are examined and Footnote 1 on page 126 is interesting for its reference to the Kemari Project and Project Remus which are projects to add hardware redundancy to Xen.

Tools and techniques for the measurement of Xen performance are shown in Chapter 10, which walks the reader through basic usage of well-known tools such as Bonnie++, httperf, UnixBench, and others. More importantly for the Xen admin is the discussion of Xen-aware profilers like Xenoprof which is "a version of OProfile that has been extended to work as a system-wide profiling tool under Xen..." (151).

Chapter 11 covers the Citrix XenServer, which is the enterprise-grade commercial Xen product from Citrix. The authors summarized it best in the review of Chapter 11: "Can Citrix's product replace the open source Xen? As always, the answer is maybe. It offers significant improvements in management and some interesting new capabilities, but that's balanced against the substantial cost and annoying limitations" (174).

Chapter 12 begins the discussion of Hardware Virtual Machines (HVMs), which are virtualization extensions that enable "an unmodified operating system [to run] as a domU" (176). This means the ability to run an unmodified version of Microsoft Windows as a guest OS within a Xen environment. The HVM discussion in Chapter 12 leads nicely into Chapter 13, "Xen and Windows".

The main chapters of the book end with Chapter 14, "Tips", and Chapter 15, "Troubleshooting". Both chapters draw on the experience of the authors and provide value to the book for their recommendations. Though the tool of choice for troubleshooting is the nearest Google search box, it's still helpful to glance over the content in the Troubleshooting chapter if for no other reason than to maybe remember that it's there when you receive the dreaded "Error: DestroyDevice() takes exactly 3 arguments" error.

The Book of Xen is almost certainly a time-saver for anyone looking to implement Xen or virtualization with Linux. The back cover states "The Complete Guide to Virtualization with Xen". The book lives up to that statement and more.

You can purchase The Book of Xen: A Practical Guide for the System Administrator from amazon.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Sorry! There are no comments related to the filter you selected.

They missed their chance! (2, Interesting)

Locke2005 (849178) | more than 4 years ago | (#30433420)

They should have named this book The Zen of Xen.

Re:They missed their chance! (2, Informative)

Itninja (937614) | more than 4 years ago | (#30433472)

I think that's been done...about 20,000 times [google.com] .

Re:They missed their chance! (1)

Locke2005 (849178) | more than 4 years ago | (#30433560)

Frightening... not only has google already indexed the parent, but it's number 2 in the rankings for the linked search!

Re:They missed their chance! (1)

topcoder (1662257) | more than 4 years ago | (#30433670)

You are right, i am trapped in an infinite loop of 2 links!!!

Re:They missed their chance! (1)

Tubal-Cain (1289912) | more than 4 years ago | (#30433780)

No. 1 now...

Re:They missed their chance! (1)

jasonmicron (807603) | more than 4 years ago | (#30433782)

Brilliant, really.

Re:They missed their chance! (1)

gmuslera (3436) | more than 4 years ago | (#30433564)

The alternate name, "Xen and the art of (something)" had even more hits.

Re:They missed their chance! (2, Informative)

TheRaven64 (641858) | more than 4 years ago | (#30446384)

Not surprising, because that was the title of the first research paper that presented Xen. Xen and the Art of Virtualization is the paper that describes Xen 1.0.

We felt that Persig puns had been done. (1)

LukeCrawford (918758) | more than 4 years ago | (#30435470)

not that I have anything against Persig, but xen/zen puns are pretty worn out. We had a bit of a scare when we noticed that the proof for the rear cover said "Xen and the art of virtualization" - I mean, Our "no persig puns" rule aside, that's what the Cambridge folks called one of the first papers on Xen, so there would have been copyright issues. No starch caught the mistake before it hit paper, though.

Re:We felt that Persig puns had been done. (1)

Ethanol-fueled (1125189) | more than 4 years ago | (#30436150)

Zen and the art of Motorcycle Maintenance totally sucked in the same way Steppenwolf sucked, and not because it is "deep" and difficult to understand. Those books sucked because they are like Tori Amos songs - a bunch of "random" self-indulgent garbage which could be applied and interpreted and to mean anything. At least David Lynch's random self-indulgent garbage is somewhat entertaining.

Both of those books should be modded -1 overrated.

Zen and the art of Motorcycle Maintenance (1)

LukeCrawford (918758) | more than 4 years ago | (#30436390)

Was recommended to me by someone I respect quite a lot, someone who isn't known for being 'fuzzy headed'

But yeah, I read the book and I didn't get it, either.

Re:Zen and the art of Motorcycle Maintenance (2, Insightful)

Ethanol-fueled (1125189) | more than 4 years ago | (#30436618)

You misunderstood me, I did get it. And it was barely a step up from from, "Confucius say, man who stand on toilet is high on pot". I'm sure it appears to be "deep" to the geek who's never read any real literature. It's the Twilight of philosophy.

what philosophy do you like? (1)

LukeCrawford (918758) | more than 4 years ago | (#30437538)

Personally, after the age of 25 or so, I started to agree more and more with Paul Graham [paulgraham.com] on the subject

Re:Zen and the art of Motorcycle Maintenance (1)

Degrees (220395) | more than 4 years ago | (#30444252)

There was one contemplation point in the book that was outside the norm of what you would learn in school: "what is quality?" (define quality).

Prior to reading the book, you probably could not have answered that question. After reading the book, you can. It ties into the opening line of the book: "And what is good, Phaedrus,
and what is not good -- Need we ask anyone to tell us these things?"

Outside of that, it made for an interesting drama.

Re:They missed their chance! (0)

Anonymous Coward | more than 4 years ago | (#30435660)

XENU should have been its name so that it would interest Tom Cruise (Mapother)

Blakes 7 (0)

Anonymous Coward | more than 4 years ago | (#30433458)

Wasn't Xen the computer of the Liberator?

Re:Blakes 7 (1)

MRe_nl (306212) | more than 4 years ago | (#30436784)

That was just "Zen" IIRC, and joined later by the portable "ORAC".

Book of Xen (3, Funny)

Spatial (1235392) | more than 4 years ago | (#30433478)

Chapter 1: How to crouch-jump.
Chapter 2: Platforms, the silent killer.
Chapter 3: Giant psychic fetuses and you.

Re:Book of Xen (2, Funny)

HisMother (413313) | more than 4 years ago | (#30433728)

Hey, did you see a man with a briefcase?

ethics (0)

Anonymous Coward | more than 4 years ago | (#30433530)

In the interest of ethics, you should also now not work for this publisher for 2 years.

Does XEN have a future? (1)

Anonymous Coward | more than 4 years ago | (#30433646)

Haven't the kernel developers basically given xen the cold shoulder, preferring now to dedicate their efforts to kvm, the virtualization solution which is now part of the mainline kernel? The problem, from what I recall from reading some threads on the subject a while back was that xen needs to keep pace w/ kernel driver development work, which is an extra effort for the developers. It's easier for them with kvm since it's part of their dev tree. Right now I think Ubuntu is the only big distributer shipping kvm; Redhat et al are doing xen .. I think (corrections please). On that count, xen is doing well. But moving forward, it seems like they have to do a lot of extra work to track kernel development, which will be made harder by virtue of the fact that key kernel developers may not feel like devoting time to what they feel is a duplicate effort.

Re:Does XEN have a future? (0)

dikdik (1696426) | more than 4 years ago | (#30433672)

Xen and KVM are completely different types of virtualization solution. The supposed rivalry between the two is largely bad journalism, not rooted in anything to do with the platforms themselves. If you want to run a single physical computer with multiple operating system instances, such as replacing a bank of servers with a single machine, Xen is your guy. If you want to run VMs under Linux, KVM is your friend. Conflating the two is like comparing... well, to use a car analogy, for this is Slashdot, a railroad with a tractor trailer.

Re:Does XEN have a future? (3, Insightful)

dikdik (1696426) | more than 4 years ago | (#30433726)

Xen and KVM are completely different types of virtualization solution.

Well yes they do do things differently, but KVM does it better and simpler by just running on Linux as the base system hypervisor. From a maintenance point of view things get far simpler, as the OP said.

If you want to run a single physical computer with multiple operating system instances, such as replacing a bank of servers with a single machine, Xen is your guy. If you want to run VMs under Linux, KVM is your friend.

That statement is just, well, daft. You're implying that Xen can't run VMs under Linux but KVM can, or Xen can run VMs on systems other than Linux or something that KVM can't do? They're both Linux only at this point, and Xen effectively runs a forked version of Linux because it isn't, and won't be, upstream.

Re:Does XEN have a future? (1)

palegray.net (1195047) | more than 4 years ago | (#30433900)

They're both Linux only at this point, and Xen effectively runs a forked version of Linux because it isn't, and won't be, upstream.

This is false. You can actually run various BSDs under Xen, and you can run Windows too.

Re:Does XEN have a future? (2, Informative)

reset_button (903303) | more than 4 years ago | (#30434256)

They're both Linux only at this point, and Xen effectively runs a forked version of Linux because it isn't, and won't be, upstream.

This is false. You can actually run various BSDs under Xen, and you can run Windows too.

I believe he meant that the hypervisor itself is a forked version of Linux. Sure, you can run any guest OS you want on both Xen and KVM.

Re:Does XEN have a future? (1)

Galactic Dominator (944134) | more than 4 years ago | (#30434572)

Either way it's still wrong. Dom0 isn't limited to linux, opensolaris and netbsd are also capable of running that hypervisor. Amazon has also been prodding the FreeBSD Foundation to finish their work in Dom0 as well.

Re:Does XEN have a future? (1)

LukeCrawford (918758) | more than 4 years ago | (#30435498)

we've got a chaper on running OpenSolaris and NetBSD Dom0s in the book. We got our first taste of Xen, in fact, on NetBSD before moving to Linux Dom0s (we're talking about moving back, now that NetBSD 5 is out with x86_64 and i386PAE support)

Re:Does XEN have a future? (1)

LukeCrawford (918758) | more than 4 years ago | (#30435514)

"The hypervisor is linux" is the idea behind KVM. The Xen hypervisor is most certainly not Linux.

Re:Does XEN have a future? (3, Informative)

Lemming Mark (849014) | more than 4 years ago | (#30434132)

Disclaimer: I've been a Xen developer in the past and I'm currently obliged to work on Xen-related stuff for my degree.

Minor clarification: Xen support for running as a domU is upstream and there are plans to get dom0 support upstream as well. If getting dom0 support upstream doesn't happen, at least (hopefully) they might be able to base their dom0 kernel on a patchset on top of the kernel.org domU support rather than the horrible forked business they used for a long time :-S

Re:Does XEN have a future? (1)

nwmcsween (1589419) | more than 4 years ago | (#30435926)

Isn't dom0 itself a hypervisor kernel? If so I highly doubt Linux will integrate a kernel within itself.

Re:Does XEN have a future? (1)

Lemming Mark (849014) | more than 4 years ago | (#30436492)

Isn't dom0 itself a hypervisor kernel? If so I highly doubt Linux will integrate a kernel within itself.

Dom0 isn't actually a hypervisor, no. Xen itself is the true hypervisor on the system and it controls access to the CPU, system memory and various other very low level concerns. Any OS that runs on a Xen system must be a virtual machine of some kind.

Dom0 is a special virtual machine that happens to be allowed to do privileged stuff like accessing the real hardware on the system and managing other domains. Basically you get a dom0-capable kernel by starting from a paravirtualised domU kernel and adding various stuff to support these privileged operations.

The various dom0 kernel patches to mainline extend the existing domU support in various ways so that these features are supported. You still need to boot on top of Xen but the advantage to getting this in mainline (and using paravirt_ops) would be that you can build a single kernel imagine that runs paravirt as dom0 or domU or boots natively or whatever.

There has been some objection to adding dom0 support, on the grounds that KVM is now the "Linux way". However, it would be useful to people and would be some way short of adding a second full hypervisor - Xen itself would continue to be a separate codebase. It'd be more a case of adding another (strange) kind of specialised "machine" that Linux is capable of booting on.

NetBSD has included support for booting as both dom0 and domU for some time now (the original NetBSD Xen support was written by a guy who later became a Xen developer, it's since been taken over by someone else I think). OpenSolaris has dom0 and domU support also, for Sun's own hypervisor product. FreeBSD 8 has added Xen domU support but sadly dom0 support is still not available.

Re:Does XEN have a future? (0)

Anonymous Coward | more than 4 years ago | (#30434454)

Did you just reply to and also disagree with yourself? Or is slashdot wigging out...? "dikdik (1696426)" and "dikdik (1696426)"

Re:Does XEN have a future? (1)

Ethanol-fueled (1125189) | more than 4 years ago | (#30436388)

Hahahahaha! Toady sockpuppet detected. Thanks for pointing that out.

Maybe both diks are twin brothers who frequently disagree with each other but share the same account. Or ol' dikdik may have disassociative identity disorder.

Re:Does XEN have a future? (0)

LukeCrawford (918758) | more than 4 years ago | (#30435648)

If you want to run a single physical computer with multiple operating system instances, such as replacing a bank of servers with a single machine, Xen is your guy. If you want to run VMs under Linux, KVM is your friend.

That statement is just, well, daft. You're implying that Xen can't run VMs under Linux but KVM can, or Xen can run VMs on systems other than Linux or something that KVM can't do? They're both Linux only at this point, and Xen effectively runs a forked version of Linux because it isn't, and won't be, upstream.

Xen and KVM can do similar things, yes, however, last time I evaluated it (which was about a year ago, and things are changing fast) KVM was wholly unsuitable for production use. The stability and performance had a long way to go before it came anywhere near Xen.

On the other hand, KVM is much easier if what you really need is an accelerated qemu to test something real quick. setting up Xen on the desktop is a huge pain in the ass.

From a service provider perspective, all the parts of xen that need to be upstream are upstream. Guest support has been upstream for some time now. I'm fine running a funny Linux kernel (or a funny NetBSD or OpenSolaris kernel) in the dom0.

Re:Does XEN have a future? (1)

Eunuchswear (210685) | more than 4 years ago | (#30437126)

I'm fine running a funny Linux kernel (or a funny NetBSD or OpenSolaris kernel) in the dom0.

You must like running dodgy old software. The most recent Xen dom0 that's half-way stable I can find (Debian Lenny) is borderline for my needs (dodgy multipath, creeky mdadm, only 90% ready for recentish hardware (Dell R710)...).

I avoid the Debian port of the SUSE port (1)

LukeCrawford (918758) | more than 4 years ago | (#30437312)

of the xen stuff- I've had nothing but trouble with it in testing, and looking, it seems that the debian people aren't particularly interested in helping if you have problems, and if you ask the xen mailing lists, they tell you to ask the debian people. If you want stability, you have to deal with the 2.6.18.8-xen kernel distributed by xen.org, or the 2.6.18-patchedtohellandback kernel distributed by RedHat. (the Suse kernel might be stable, I haven't tried it, but the debian port of the suse kernel that is 2.6.27, man, that sucks. Some people say this is because they took the initial patch and have not been good about importing fixes that suse did.)

I've used both those Linux kernels (the xen.org kernel and the RHEL 5 kernel) quite a lot; my experience has been that the RHEL kernel is slightly less stable, while the xen.org kernel has serious driver deficiencies.

For my current production boxes, I'm using the xen.org 2.6.18 system, and I just buy the exact same hardware every time. (Yay for cheap sata_sil cards!)

Another option, of course, is NetBSD5. After starting on NetBSD 3, and switching to linux for pae and x86_64 support, I'm seriously considering switching back, now that NetBSD5 is starting to look good.

Re:Does XEN have a future? (1)

Eunuchswear (210685) | more than 4 years ago | (#30437046)

but KVM does it better and simpler by just running on Linux as the base system hypervisor.

No, KVM des it worse and simpler - Xen is faster than KVM on the same hardware (about 10% for disk I/O comparing Xen paravirtualisation with KVM virtio-blk).

Re:Does XEN have a future? (1)

TheRaven64 (641858) | more than 4 years ago | (#30446800)

Well yes they do do things differently, but KVM does it better and simpler by just running on Linux as the base system hypervisor. From a maintenance point of view things get far simpler, as the OP said.

This is quite a misleading assertion. I wrote a book about the Xen internals [amazon.co.uk] , so I'm somewhat biased, but you make it sound like KVM is a much simpler approach than Xen. There are a few issues with this.

Xen is a hypervisor. It runs in privileged mode (ring 0 on x86) and is very small. Linux is a kernel ('supervisor') that runs in privileged mode and is massively more complicated than Xen. The confusion comes from the fact that Xen delegates a lot of things to the domain 0 guest. Rather than supporting device drivers itself, the Xen hypervisor allows a guest to connect directly to the hardware and export virtual devices to other guests. If your hardware supports something like Intel's VT-d then you can put each device driver in a different privileged guest kernel and have them completely isolated form each other and from unprivileged guests. With KVM, all of the device drivers run with kernel privileges.

Xen also delegates the management code to a guest. This is usually the same domain 0 guest that runs the device drivers, but it doesn't have to be (it does initially for bootstrapping, but not necessarily after that).

It's also worth noting that Xen does not require Linux. You can use Xen as a hypervisor with NetBSD or Solaris as the domain 0 guest. Solaris has some nice features for offline crash debugging (it dumps the crash data into the swap partition and read them back on reboot into a core file) which have been ported to work with Xen. You can also use ZFS ZVOLs with Xen on a Solaris dom0, so you can create new VMs trivially just by cloning an existing volume and starting a new VM instance connected to it.

I'm not sure what KVM support is like on non-x86 platforms, but Xen also works on ARM, PowerPC and IA64. I was at the XenSummit in 2007 and Samsung had a nice demo of Xen on an ARM-based handheld booting two operating systems and switching between them so you could completely sandbox untrusted code. They tweaked the driver model slightly so some devices (notably the screen and stylus) were assigned exclusively to one VM at a time, rather than being multiplexed like the network and storage devices. There's also been a lot of work done on migrating between virtualized and emulated environments with Xen, which is great for debugging. I'm not sure if this has hit the main tree yet, but it's pretty neat being able to move a guest into QEMU and step instructions, revert to an earlier snapshot, and then move it back into a VM once you've fixed a bug.

KVM seems like a continuation of the Linux philosophy: do everything, but don't do anything well.

Re:Does XEN have a future? (4, Informative)

MightyMartian (840721) | more than 4 years ago | (#30433866)

You make KVM sound like it's nothing more than a desktop virtualization package, as if somehow it's less of a virtualization solution than Xen, and that's just bullshit.

I'm running KVM on Ubuntu 9.10 right now with three guests; Windows Server 2003, an Ubuntu 8.04 install running as a file server and a Debian Lenny install for my webserver. With the newer versions of KVM, I should be able to get a SCSI card running on a guest (paravirtualization), so I'll be giving my backup server a shot as a guest as well.

I'm not trying to claim KVM is better than Xen, but the characterization of KVM that you give might have applied a couple of years ago, but it's just uninformed crapola now.

Re:Does XEN have a future? (0)

Anonymous Coward | more than 4 years ago | (#30434624)

With the newer versions of KVM, I should be able to get a SCSI card running on a guest (paravirtualization), so I'll be giving my backup server a shot as a guest as well.

That is not paravirtualization (unless you start calling any kind of abstraction "paravirtualization").

KVM does not do paravirtualization, it virtualizes a full x86 processor (with all its overhead).

Re:Does XEN have a future? (2, Insightful)

reset_button (903303) | more than 4 years ago | (#30434762)

KVM does not do paravirtualization, it virtualizes a full x86 processor (with all its overhead)

KVM does not need to do para-virtualization, but it can do para-virtualized I/O to get better performance (see virtio [linux-kvm.org] ).

Re:Does XEN have a future? (0)

Anonymous Coward | more than 4 years ago | (#30435272)

Which one should I prefer? I need to make a virtualisation decision soon but what are the pros and cons with KVM and Xen? I will be using debian stable. Live migration is a must as is hardware supported virtualisation.

Re:Does XEN have a future? (1)

Eunuchswear (210685) | more than 4 years ago | (#30437214)

I will be using debian stable. Live migration is a must as is hardware supported virtualisation.

If by stable you mean Lenny then Xen is the best - Debian's Xen works great on Lenny. If you want to use Squeeze then you have a problem, dom0 doesn't work yet, but it's been promised. After Squeeze you're screwed.

Personally? I would not use the debian Dom0 (1)

LukeCrawford (918758) | more than 4 years ago | (#30437384)

DomU works, and should continue to work for the foreseeable future; DomU paravirt ops is upstream, so unless debian goes through extra effort to break xen DomU support, it should Just Work.

Re:Does XEN have a future? (0)

Anonymous Coward | more than 4 years ago | (#30435430)

I've encountered two things that did not work under KVM, but do under VMware and the real-deal (not sure about under Xen):

1. Comodo Firewall on a WinXP SP3 VM. Comodo will crash within one minute of login.

2. Activation *cough* patches *cough* for Windows 7 VM. Win7 itself works fine, but with the bootloader installed, it's a black-screen on bootup.

KVM's great, but there's still a few compatibility issues here and there to work out.

Re:Does XEN have a future? (1)

MightyMartian (840721) | more than 4 years ago | (#30438226)

It's worked for pretty much everything I've thrown at it. I haven't tried some of the more complex networking schemes, so I haven't tried building a NAT router. I might play with that at some point in the future, and toss out my existing Linux-based router. Then

Re:Does XEN have a future? (1, Interesting)

Anonymous Coward | more than 4 years ago | (#30434552)

If you want to run a single physical computer with multiple operating system instances, such as replacing a bank of servers with a single machine, Xen is your guy. If you want to run VMs under Linux, KVM is your friend. Conflating the two is like comparing... well, to use a car analogy, for this is Slashdot, a railroad with a tractor trailer.

I believe this statement is False. I replaced a bank of servers using Xen. It worked fine, but I needed better driver support and its fork of 2.6.18 or whatever wasn't cutting it.

I switched to KVM two years ago and have had better success than with XEN. On one host right now, I have 8 Windows 2k guests including a SQL Server, Exchange servers, domain controllers and web servers. There are also 4 various flavors of linux servers running.

Using KVM has been a joyous breeze compared to dealing with XEN.

Re:Does XEN have a future? (1)

ka8zrt (1380339) | more than 4 years ago | (#30435082)

One of my biggest complaints about KVM is the fact that it requires CPUs which support the HVM CPU capabilities (i.e. Intel's VT or AMD's AMD-V). For a site which has already has hundreds, if not thousands of systems running older P4/Xeon hardware, expecting them, or a small company with fewer systems to upgrade to get newer CPUs which can support KVM is not reasonable. Add this to the whole pile of fecal material surrounding the kernel team not putting the ParaVirtOps code into the kernel tree, and Xen Dom0 not being supported on things like Fedora Core since FC8, and RHEL/Centos lagging so far behind on kernels, it is enough to make even a normally mild mannered person want to Gibbs some developers, if not worse.

And yes, HVM is nice in that I could run pretty much any OS on it, with no modifications, be it the latest release of some Linux distro, NetBSD or other OS, or a legacy system running NetBSD 2.x, or (gack) WinXP or even Win95/98... but Xen could do that just as easily as KVM.

Try qemu or virtualbox (1)

Colin Smith (2679) | more than 4 years ago | (#30435566)

Not as quick as with hardware virtualisation and consumes rather too many host cycles, but kqemu in particular is pretty straightforward to set up on older CPUs with rhel or centos (just make sure the kernel module version is compatible with the emulator).

 

Personally, I don't think that is an issue (1)

LukeCrawford (918758) | more than 4 years ago | (#30435876)

if you pay for power, replacing 8 P4s with 1 dual quad-core xeon or opteron usually saves you enough in power costs to pay for the capital cost in a few months.

For me, the big differentiator is robustness and reliability. KVM will get there, I'm sure. But in the hosting space, Xen is the established tool, and KVM is the new technology. Nearly all the linux developers are rooting hard for KVM, which may mean that it will become better. Personally, I'm hedging my bets by learning both. But for now, at least, I think Xen is the best choice for the hosting provider.

Re:Does XEN have a future? (1)

dgym (584252) | more than 4 years ago | (#30435106)

Xen and KVM are completely different types of virtualization solution. The supposed rivalry between the two is largely bad journalism, not rooted in anything to do with the platforms themselves.

How is that? They both support full virtualization, and they both support paravirtualization for things like networking and disks (they even use the same protocols for this - virtio). Most of the debate I have seen regarding the two is whether it is better to have a specialized hypervisor (Xen) or use a general purpose, but heavily developed and very much optimized, hypervisor (Linux). The performance benchmarks are six of one and half a dozen of the other so this debate doesn't seem overly relevant.

If you want to run a single physical computer with multiple operating system instances, such as replacing a bank of servers with a single machine, Xen is your guy. If you want to run VMs under Linux, KVM is your friend. Conflating the two is like comparing... well, to use a car analogy, for this is Slashdot, a railroad with a tractor trailer.

I'm not sure why you think this is the case, I consider them to be pretty much equivalent. The reason we use KVM is because Xen wasn't up to date enough to run on our brand new hardware. KVM does a fine job as a Xen replacement, and we have been using it as such for years.

Re:Does XEN have a future? (2, Interesting)

reset_button (903303) | more than 4 years ago | (#30434234)

Redhat et al are doing xen .. I think (corrections please)

Correction: Red Hat aquired Qumranet, the inventors of KVM (link [redhat.com] ), so Red Hat is abandoning Xen in favor of KVM as well.

Re:Does XEN have a future? (3, Informative)

Lemming Mark (849014) | more than 4 years ago | (#30434340)

Disclaimer: I've been a Xen developer in the past and I'm currently obliged to work on Xen-related stuff for my degree. However, please take the following as my personal opinions!

Wrt keeping pace with kernel driver development: it's worth noting that most of Xen's hardware support comes from running device drivers within a privileged Linux VM (dom0). So most things supported under Linux should work under Xen, plus new drivers should also work as they're added (assuming you're running an up-to-date Linux, which as others have noted you may not be given Xen's officially supported dom0 has mostly been based on 2.6.18, though I hear rumours that proper support for running kernel.org kernels as dom0 might be coming or already available somewhere). That said, stuff like power management / CPU frequency scaling / hibernate / etc tends to need co-operation from Xen, so that can lead to some duplication and extra complexity for Xen to support (I don't know that it does support hibernate support, for instance).

I think that the kernel developers have given Xen support in mainline a fairly rough ride (though as the patches were somewhat large and unusual, that's partly understandable). The Xen developers could probably have handled the merge faster by engaging more and better with the kernel community earlier on.

Regarding distro support - RedHat made Xen a major feature, with good Xen support in RHEL5 (and later support for RHEL4 running as a PV Xen guest) and Fedora. However, relatively early on they started isolating their command-line tools and GUI from the specifics of Xen through an abstraction library called libvirt, a command shell called virsh and a gui called virt-manager. They are now going with KVM as their primary virtualisation feature for the future, with the same UI provided. Obviously they'll have to keep supporting Xen in RHEL due to long term support guarantees. Fedora hasn't supported running as dom0 for a couple of releases as they've decided to stop porting the Xen patches and wait for kernel.org support (they did provide some development and impetus to the mainlining effort, to help this goal along). RedHat has bought the company which wrote KVM and also acquired their SPICE virtual desktop protocol and additions to KVM / Qemu to support this. SPICE is currently most useful for Windows guests from what I've read but it can do some pretty impressive stuff.

SuSE are still quite keen on Xen AFAIK and Xen is certainly still an OSS project which continues to be developed and enhanced. That said, XenSource (which is now owned by Citrix) isn't really in the business of providing virtualisation for Linux distros (although the open source Xen can do that), so much as in selling "appliance-style" virtualisation software a la VMware ESX. They want to be attractive as a commercial virtualisation platform for whatever is expected to be popular in data centres and they're not so worried about being a de-facto standard in Open Source circles.

Re:Does XEN have a future? (0)

Anonymous Coward | more than 4 years ago | (#30482294)

They want to be attractive as a commercial virtualisation platform for whatever is expected to be popular in data centres and they're not so worried about being a de-facto standard in Open Source circles.

Which seems like a foolish choice.

The window of opportunity for Xen is closing to be the popular choice in data centers. I think once (if... I think RH is still on the fence about KVM vs Xen) RHEL starts recommending KVM over Xen that Xen will be pretty much pushed out of the data center. There are a lot of sysadmins out there who equate RHEL with data center.

Oh, you want a nice Xen environment? (0, Offtopic)

palegray.net (1195047) | more than 4 years ago | (#30433662)

Buy a Linode [linode.com] ;).

Sorry, I just couldn't resist.

Re:Oh, you want a nice Xen environment? (1)

LukeCrawford (918758) | more than 4 years ago | (#30437708)

haha. Linode is one of the better providers. they are cheaper than slicehost, but more expensive than I am. According to the latest benchmarks I've seen, though, they beat us all (slicehost, ec2 'small', prgmr.com) in terms of CPU power. (though I'm pretty up front about the fact that I optimize for cheap ram over all else, so the results are unsurprising)

Re:Oh, you want a nice Xen environment? (1)

palegray.net (1195047) | more than 4 years ago | (#30439940)

Well, fancy meeting you here :). I'd go so far as to say we're pursuing somewhat different markets, too. I've certainly heard folks who put a premium on price speak favorably of your offerings. Anyhow, back to hacking on stuff. Have a good one!

Isn't everyone like just using KVM? (2, Interesting)

Colin Smith (2679) | more than 4 years ago | (#30433696)

Y'know, there by default vs reinstalling your kernel and patching everything every release etc.
 

Re:Isn't everyone like just using KVM? (0)

Anonymous Coward | more than 4 years ago | (#30434042)

Every release isn't a problem with Xen. Releases have mostly stopped happening, so you get the joy of being stuck on a kernel that was outdated three years ago for your dom0.

No wonder Red Hat have all but abandoned it. It's a code-maintenance nightmare.

Re:Isn't everyone like just using KVM? (2, Informative)

jsolan (1014825) | more than 4 years ago | (#30435630)

No, those of use with older servers (without vmx/svm processors) are not. The machines work just fine and XEN allows us to paravirtualize several linux machines on each one. Unfortunately RedHat has lost sight of those customers. As far as I'm aware KVM still requires the intel vt/amd-v technologies to create any guests, which XEN only requires for HVM guests (i.e. windows).

Re:Isn't everyone like just using KVM? (1)

digitalhermit (113459) | more than 4 years ago | (#30436656)

RedHat is supporting Xen for the lifetime of the RHEL5 spin, which is at least a couple years from now. I think you have until 2014 until they stop supporting Xen.

I was skeptical of the move to KVM but I think it's a good technology. It has some very interesting features that Xen does not, including KSM (kernel same-page merging) and better memory balooning. Plus it makes much better sense from a support standpoint. Xen is owned, at the top, from a proprietary company (Cisco). It was only recently (Oct 2009) that it was completely opened up (GPL). But the fact that KVM is in the mainline kernel means that it's easier to support on RedHat kernels, and that alone may make it worthwhile.

As to the vt/amd-v requirements, yeah, that's needed. The number of systems without chip-level virtualization is falling though.

Re:Isn't everyone like just using KVM? (2, Informative)

qnetter (312322) | more than 4 years ago | (#30441738)

Wow - how many inaccuracies can you pack into one comment?!

Xen itself - including not just the hypervisor but the kernel code needed for dom0 and for paravirtualized domU - is GPL licensed, and always has been. What Citrix (NOT Cisco) recently open-sourced was the control stack used in he commercial XenServer. There has always been an open-source control stack -- it has been possible to run a Xen system entirely using GPL licensed code. The only change in October 2009 was to make the management APIs compatible between the commercial and open-source offerings.

(And the "owned by Citrix" part is open to debate too. XenSource -- a company -- was bought by Citrix -- not Xen -- a code base. The licensing status of the code is the same as it ever was. Its direction is driven by an advisory board that includes representatives of Citrix and lots of other companies.)

prgmr.com (0, Offtopic)

selfabuse (681350) | more than 4 years ago | (#30433880)

Luke Crawford is great - I used to have a virtual server hosted with his company, prgmr.com [prgmr.com] . You can feel pretty confident in your xen machine when the guy who maintains it literally wrote the book on it.

Re:prgmr.com (1)

stevey (64018) | more than 4 years ago | (#30433942)

By contrast I've never heard of this book, or this guy, but I wrote xen-tools & xen-shell...

Re:prgmr.com (0)

Anonymous Coward | more than 4 years ago | (#30434166)

What the fuck are those?

Re:prgmr.com (0)

Anonymous Coward | more than 4 years ago | (#30434294)

Read the review ..

You have a short memory (1)

LukeCrawford (918758) | more than 4 years ago | (#30435684)

I'm pretty sure I've emailed you to ask dumb questions several times, and you answered. :)

Re:You have a short memory (1)

stevey (64018) | more than 4 years ago | (#30438546)

I'm sorry for my early morning terseness, then!

Re:prgmr.com (1)

pwnies (1034518) | more than 4 years ago | (#30447280)

Why is this modded off topic? Luke Crawford, the same guy who wrote this book, runs prgmr.com - a xen based vps host. If anything it shows that the author is competent. Mod parent informative.

Oh my god, that's cute! (1)

zish (174783) | more than 4 years ago | (#30434134)

I cannot believe that I'm the first to post about hot cute the cover is. At least anyone will have to agree that the subjects would make for interesting sushi.

Re:Oh my god, that's cute! (1)

armanox (826486) | more than 4 years ago | (#30434496)

I don't think I'd try the Windows fish for sushi. (Clever all the OS fish in one tank)

The review needs ... google (2, Insightful)

vlm (69642) | more than 4 years ago | (#30434560)

The review needs to list, what the book has, that a google search will not find for free.

The disadvantage of the book, is the review makes it sound like an edited, slightly out of date collection of FAQ printouts, and it costs money.

The advantage of the book, is ... Come on, give me a reason to buy the book. I got money burning a hole in my pocket here.

Here are reasons why I'd buy a book that covers a "google-y" topic. First of all, if the $$$ went to a foundation that paid the program authors, or some charity (EFF?). Maybe if it had diagrams and flowcharts or other graphical aids you can't find via google. Maybe if it had interesting exclusive content, behind the scenes interviews with the authors, candid explanations of why and how it was designed the way it was, that simply cannot be found on the internet. Maybe if it had workbook like qualities, like planning worksheets to plan and organize your deployment. Maybe if it had textbook like qualities like questions at the end of each chapter with an answer key at the end. Maybe if it was an artistically beautiful collection of "stuff" like XKCD cartoons, FAQs, stories, pictures, poems, all tied to a common theme in an aesthetically pleasing manner.

Thats what I want to see in a slashdot review of a "google-y" topic.

Re:The review needs ... google (0)

Anonymous Coward | more than 4 years ago | (#30435056)

Add to this that the state of virtualiztion is changing by the day. Sure, virtualizing a server can be done relatively easily but I'm talking about roaming profiles, moving between virtual and physical, managing massive deployments, moving resources around, using templates, virtualizing applications, dektops etc. Man, even something as simple a what protocol to use attaching to the virtual sever is changing by the day, ICS, RDP (which version) PC0IP. How about USB redirection, multiple desktops, thin and thick provisioning, two way authentication, security gateways, video delivery blah blah blah. The state of virtualization above the simple small office moving a few servers to virtual has a while until the dust settles. If your just moving a few servers, you don't need this book, if you are moving hundreds of servers and possibly integrating some desktop virtualizations, this book will be outdated as soon as it leaves the publisher.

half what you are talking about is (1)

LukeCrawford (918758) | more than 4 years ago | (#30437942)

'desktop virtualization' which is another animal entirely. (personally, I think desktop virtualization will suffer the same fate of other recent 'thin client' schemes, but really, it's not my area, so that's just wild speculation.)

Provisioning and mass deployments is something I go into a little bit... but certainly isn't the focus of the book. I firmly believe that it is madness to treat physical and virtual servers differently; you want to use one tool for both. And right now, the best one tool is cobbler/koan.

If I may (1)

LukeCrawford (918758) | more than 4 years ago | (#30435952)

books can't really compete as references anymore, I don't think. The advantage of a book is that it's easier to sit down and read a book cover to cover than to figure out what you need to look up in order to get an overview of a technology.

Re:The review needs ... google (1)

drsmithy (35869) | more than 4 years ago | (#30440332)

The review needs to list, what the book has, that a google search will not find for free.

Off the top of my head, I'd go with coherency, structure, and a lack of insults.

Re:The review needs ... google (1)

vlm (69642) | more than 4 years ago | (#30442836)

The review needs to list, what the book has, that a google search will not find for free.

Off the top of my head, I'd go with coherency, structure, and a lack of insults.

Sounds just like a FAQ... which google will find for free

Google Xen FAQ first link for me was

http://wiki.xensource.com/xenwiki/XenFaq [xensource.com]

Good coherency, although it was not, it looks like it was written in one sitting by one guy, or at least edited to look that way.

Good structure, its an outline, and the outline even looks well designed.

Lack of insults, well, its no mailing list. And for $20 or whatever the book costs, I'll glance over a mailing list post calling some dude a noob, or whatever.

XEN has a way to go yet (1)

munrom (853142) | more than 4 years ago | (#30435236)

XEN isn't bad, runs ok unless it's a windows guest. What it lacks it good administrative interface. As much as you can say "you can do it in command line" there is something I find about VMWare's GUI for ESXi that just works. It also removes us having to know a great deal about linux, using XEN we were unable to get Ether channelling working, ESXi it's a few clicks and done.

Re:XEN has a way to go yet (1)

LukeCrawford (918758) | more than 4 years ago | (#30435908)

hah. being familiar with linux, and not so familiar with Windows, I had the opposite problem. Xen networking is... Linux networking. (the big problem is that the Dom0 kernel is crusty and ancient; something that should be remedied with Xen 4, which should be out Real Soon Now.)

Re:XEN has a way to go yet (1)

TheRaven64 (641858) | more than 4 years ago | (#30447012)

Did you talk to anyone at XenSource (well, the bit of Citrix that used to be XenSource)? Last time I chatted to Simon Crosby, people running Windows guests on top of Xen accounted for around 90% of their market, and they had a custom WinNT HAL for Xen.

Re:XEN has a way to go yet (1)

LukeCrawford (918758) | more than 4 years ago | (#30453404)

yeah, they went over the 'citrix xen' chapter.

But what I was trying to say is that networking under xen is pretty easy for me, as it's just Linux stuff. The times I've had to deal with vmware for clients, I've been frustrated trying to do what I wanted to do with their interface, something obviously designed for windows users. I'm not saying the vmware interface is bad; just that it was designed for a windows user.

Something completely off topic (1)

jandersen (462034) | more than 4 years ago | (#30442058)

- or nearly:

This book title is a wordplay on "Zen", obviously; which leads me, once again, to wonder why the hell English speakers think that "X" is pronounced "Z"? Is it just mental laziness? The depravity that inevitably comes with the filthy excesses of Capitalism? A conspiracy by the religious right?

In my experience "x" is pronounced "ks", as demonstrated in a word like "sex" - you don't pronounce it "sez" unless you've had a few pints more than what is good for you, in which case sex is probably irrelevant anyway. It doesn't take much effort to say "Ksen" (like it is spelled, not "ker-SEN", which seems like a more reasonable pronunciation of "Xen".

*huff, huff* Right, I feel much better now. Thank you for listening.

Re:Something completely off topic (1)

doobie22 (970556) | more than 4 years ago | (#30445076)

Describing it as laziness could be considered a bit of a faux pas.

Re:Something completely off topic (1)

jandersen (462034) | more than 4 years ago | (#30455190)

Describing it as laziness could be considered a bit of a faux pas.

Hey, I hope that the general tone of my post would be a hint of its entirely humorous intent. I have never heard any good explanation of why "X" ahould be pronounced "Z" when it stands first in syllable, that's all; being a Danish speaker, I am no stranger to pronouncing chains of consonants, so it puzzles me that Enhlish speakers seem to go to such lengths to avoid it, by inserting vowels or even changing the consonants completely.

Re:Something completely off topic (1)

doobie22 (970556) | more than 4 years ago | (#30455432)

I suppose my use of the word faux escaped your own humor.

Link to Errata (0)

Anonymous Coward | more than 4 years ago | (#30448598)

There's a link to the errata at the bottom of http://www.nostarch.com/xen.htm

Errata available! (1)

takempa (1701218) | more than 4 years ago | (#30452388)

See the bottom of No Starch's page on "The Book of Xen for a link.
Check for New Comments
Slashdot Login

Need an Account?

Forgot your password?