goombah99 writes "After snapping up virtualization company InnoTek at the beginning of the year, Sun has recently released VirtualBox as a fully functional and highly polished free GPL open source x86 Virtual Machine. It can host 32- or 64-bit Linux, Windows XP Vista and 98, OpenSolaris and DOS. It runs on Mac OS X, Windows, and Unix platforms. The download is just 27MB. A review of it on MacWorld, showing HD movies playing inside windows XP on a mac, demonstrates performance visually indistinguishable from VMware. Like its competition, it can run other OSes in rootless, rooted, or seamless modes display modes (where all the applications have their windows mixed at the same time). Each VM instance can only run single core (though I/O is multi-core), and it does not yet support advanced windows graphics libraries however, so some gamers may be disappointed. Slashdot discussedthe InnoTek acquisition earlier.
True. VirtualBox was GPL2'd in January 2007, Sun acquired InnoTek in February 2008 (source [wikipedia.org]). So it was actually InnoTek, not Sun, who released VirtualBox into the wild in the first place.:-)
Wow, I had never heard of this before, sounds good! I had been using Microsoft's software, which gives me a "Processor error" when I try to boot Ubuntu, perhaps VirtualBox will have better luck.
Sun just get better and better, I'm slowly finding myself using more and more of their software, and most of it is excellent.
In my experience, I've actually found VirtualBox to be much faster than VMware, and coupled with the far less demanding system requirements (at least for the VM software itself, it doesn't do much to reduce guest sys requirements, of course:P), I haven't used VMware for over a year and half now.
I had nothing but problems with it when I was testing it a couple of months ago. I couldn't get the networking to work in NAT mode, and bridging mode on a laptop ain't always the best idea. Maybe I'll give it another shot.
by Anonymous Coward
on Wednesday July 30 2008, @05:47PM (#24409823)
I couldn't get the networking to work in NAT mode, and bridging mode on a laptop ain't always the best idea.
There was a nice bug in 1.6.0 that severely hindered networking, it has been fixed in 1.6.2 though. I only had problems with bridges and tun devices, I didn't try NAT, the bug reports had windows hosts and Linux guests, my situation had Windows and Linux guests on a Linux host. To summarize the bug: networking works perfectly until you reboot the VM, then there is no working network.
I had nothing but problems with it when I was testing it a couple of months ago. I couldn't get the networking to work in NAT mode, and bridging mode on a laptop ain't always the best idea. Maybe I'll give it another shot.
Getting the networking system to work is a bit of a pain, but I've only had minor difficulties when using the host interface. NAT will work, but you won't be able to ping or access any resources in your own network (which is a bad thing if you have a fileserver at home and wish to access it on a VM). There are, however, a few tutorials that can help you get started with bridging your network for Windows hosts [wordpress.com] or a variety of Linux [gentoo-wiki.com] hosts [ubuntu.com].
FreeBSD is the only guest OS I've had difficulties with (even MSDOS will work, but it requires some additions to prevent it from eating up your cycles like crazy--FreeDOS plays nicely, though). I could only ever get the NAT-based networking to work and even then it would freeze whenever IO operations peaked.
Take a look at some of those articles, and you might be able to get networking up and running in VirtualBox! I have to say, for something of a FOSS offering, it's really nice.
hookups to scanners and printers. Except for OpenSolaris and W98SE (explicitly not supported), I've had working scanner and printer support on every OS (Kubuntu/Ubuntu V7 and V8, OpenSuse11, XP) I've tried Sun Virtualbox v1.6.2 with. I've been using it to review operating systems for publication. If I'd had a scanner and printer that worked with OpenSolaris, I think it would have worked just fine there, too.
Linux webcam support is problematic whether you're trying to get it on a real or a virtual machine. Has your webcam worked on any Linux physical box you've tried it on?
I'm planning to replace VMware Server with VirtualBox completely on this box. (Debian Lenny host)
In my experience, I've actually found VirtualBox to be much faster than VMware
VMWare supports multiple CPUs (2 cores visible on Guest OS) and also supports hardware accelerated 3D. Have you tried running any 3D or multithreaded apps under VMWare and VirtualBox? I find that VMWare is quite fast if you install the VMTools in the guest OS and the integration (cross VM copy/past / drag and drop, seamless mouse pointer, etc) is quite nice.
One of the main things I like about VMWare is the "Snapshot" capability which lets you create multiple "restore" points (in an easy to use visual "tree" manager) that you can instantly return to. In fact you can have a VM automatically revert to a snapshot. Does VirtualBox have any sort of advanced snapshot management?
VMWare's real-world SMP and Multi-CPU support is actually pretty shoddy. Although it is "supported" and two virtual processors will show up on the guest, the performance is not anywhere near bare metal multi-cpu performance.
I have a client running MSSQL and Exchange as virtual servers (both multithreaded apps that make good use of multiple processors/cores) and the performance was actually better with single virtual proc than assigning multiple cores to each VM.
VMWare supports processor "pinning" however and allows you to dedicate a specific proc/core to a VM which can really boost performance.
The only problem with VMWare is that they don't support any 64bit Host OS
The documentation states otherwise [vmware.com] and you will notice the release date for VMware workstation 5.5.2 -- with 64bit Host support -- was May 2006. I have used a 64bit Host OS for VMware workstation for nearly that long.
If you are stating the free ESX Server does not support 64bit Host OS, the GSX documentation [vmware.com] from December 2005 specifically states Windows Server 2003 x64 can be used as the host OS.
Does this thing run the VM as some sort of hypervisor underneath the OS or does it piggyback the other OS's on a parent OS.
If It's a hypervisor like thing where all the OS's' are symmetric then I guess it must be getting in the way of my "normal" OS and limiting it to single core?
If it's not a hypervisor/symmetric VM and one OS is the master, Do all the OS's have full access to the hardware functions. So for example if I my mac is the master OS, and I set up a firewall set, does the windows OS have to go through the mac's firewall (and thus be protected better) or does it have direct access to the ports itself. If the latter who negotiates the conflicts when both want the CD or audio port.
Finally, are the VMs portabel from machine to machine. Or even platform to platform.
So If I create a VM on one machine, save it's state and open it on another machine, does it just run? (even the network settings?) What if the second machine was say an AMD and the first an Intel. What if the first host was a mac and the second host a linux machine?
Does this thing run the VM as some sort of hypervisor underneath the OS or does it piggyback the other OS's on a parent OS.
It's not a hypervisor. It's basically just an application that you run under your real OS.
If It's a hypervisor like thing where all the OS's' are symmetric then I guess it must be getting in the way of my "normal" OS and limiting it to single core?
If it's not a hypervisor/symmetric VM and one OS is the master, Do all the OS's have full access to the hardware functions. So for example if I my mac is the master OS, and I set up a firewall set, does the windows OS have to go through the mac's firewall (and thus be protected better) or does it have direct access to the ports itself. If the latter who negotiates the conflicts when both want the CD or audio port.
The main OS treats it as an ordinary application. The primary OS firewall will effect hosted machines, for example, and as far as audio, as long as your primary OS can deal with multiple applications playing sound, then it's a non issue. Otherwise, it happens as any other conflict would. Generally, first to open the device wins.
Finally, are the VMs portabel from machine to machine. Or even platform to platform.
So If I create a VM on one machine, save it's state and open it on another machine, does it just run? (even the network settings?) What if the second machine was say an AMD and the first an Intel. What if the first host was a mac and the second host a linux machine?
Yes, you can move the VM images around. Part of the whole point of the VM is that it is running on the virtual hardware, and doesn't have the ability to know what the physical hardware is.
by Anonymous Coward
on Wednesday July 30 2008, @04:41PM (#24409011)
The best virtualization I've found for windows hosts. Works great - I run Vista Ultimate host & Ubuntu guest in seamless mode on my laptop and everything is still fast as hell!
It looks like a viable candidate for a VM, but still a bit behind the leaders. VMWare and Parallels seem to be better choices if you can afford them, but hopefully being free as in beer and GPL will allow it to catch up rapidly and make the ongoing competition even better. If they can get 3D graphics card support running, I will be looking really hard at VirtualBox.
Parallels sold me their desktop software when I bought an Intel Mac. After repeated crashes (OS X kernel panics, not just application panics), they finally admitted that it was their fault and they hadn't read the documentation about how inter-processor interrupts were meant to work, so their kernel module crashed regularly on any Core 2 Duo machine. Their suggested fix? Buy the new version. Those pirates deserve to go out of business.
 2 Grant of license. (1) Sun grants you a personal right to install and execute the Product on a Host Computer for Personal Use or Educational Use or for Evaluation. âoePersonal Useâ requires that you use the product on the same Host Computer where you installed it yourself and that no more than one client connect to that Host Computer at a time for the purpose of displaying Guest Computers remotely. âoeEducational useâ is any use in an academic institution (schools, colleges and universities, by teachers and students). âoeEvaluationâ means testing the product for a reasonable period (that is, normally for a few weeks); after expiry of that term, you are no longer permitted to evaluate the Product.
The binaries are not Free for corporate use. The source is free (GPL) but good fucking luck compiling it on a windows machine. Maybe you could compile it on a linux machine but on windows it assumes a development environment complete with every freakin' thing under the Sun (no pun intended). I gave up after two days of trying to get it to work.
by Anonymous Coward
on Wednesday July 30 2008, @05:13PM (#24409431)
It's not odd that they want to charge for their product, it's odd that they charge for the binaries, but not the source. That's the odd thing.
It's not odd if you look at it from a contribution standpoint.
The FOSS community contributes to VirtualBox directly through help with development, testing and bug fixing on the project, as well as indirectly through their efforts on all the other FOSS projects upon which VirtualBox depends, including toolchains and mountains of utilities. Availability of source code is clearly not optional for this.
Windows binary users get a bit of a free ride on the back of all that hard work, so instead they contribute to VirtualBox by providing a bit of cash. They don't need access to the source code nor a build environment for this, and what's more, in the Windows environment it's very normal and expected to pay for your packages.
So, the VirtualBox product offering seems quite well adjusted to its two communities, and quite fair as well.
See, this is one part where a package management tool comes in handy. For example, the binaries that are provided by Sun are not free, BUT when Debian takes the GPL'd source, and makes a.deb file, it is free.
That bussiness model seems pretty fair to me. Release the code GPL, free binaries for non commerical use, and sell the binaries for corporate clients. They are essentially charging companies for the time and expertise it takes to compile it. And presumably it means they only have to offer support to paying corprorate customers.
A nice thing about that model is that it caps the price at the value added. Think sun is charging too much? compile it yourself and support it yourself. The value contained in the code itself, and value added to the code by unpaid GPL contributors, is not part of the price this way.
And that's a very nice way to make money off GPL. You're not cheating the contributors at all. And anyone can go into competition with sun for the compiling. So it comes down to charging for the value added by sun in compiling and servicing it.
Not quite the same as RedHat's model but highly simmilar
The binaries are not Free for corporate use. The source is free (GPL) but good fucking luck compiling it on a windows machine. Maybe you could compile it on a linux machine but on windows it assumes a development environment complete with every freakin' thing under the Sun (no pun intended). I gave up after two days of trying to get it to work.
Go recursive / self-hosted build. You could always set up a VirtualBox VM with the appropriate development environment to build VirtualBox:-)
I too ran into this problem where I wanted the OSE (Open Source Edition) GPL binaries on Windows. I already had Visual Studio installed, so that wasn't a big deal, but one of the requirements to build is having the MinGW g++ compiler, so now you have a situation where you need two seperate c++ compilers to compile the thing, which is kind of wierd. On top of that you need to download and install the DirectX SDK and the Windows Driver Kit, along with several open source libraries (ok, needing various library dependencies is kind of of par for the course though).
After finally getting everything downloaded and unpacked into a build tree, and getting all the command line arguments for their configure script (so it would know where to find all the libraries), the build process ran for about 1/2 hour then died with a type casting error related to the USB device driver. Now, according to the VirtualBox website, the USB wasn't even supposed to be part of the Open Source Edition (and I suspect that might be part of why I got the errors - because it was expecting it and it wasn't there).
I asked on the VirtualBox forums and developer mailing list, and after a week someone said that they got it to build by commenting out the 2 lines that generated the build error. But now I'm *very afraid*. A Debian developer who 'got rid of build errors' by commenting out 2 very critical lines of source code put hundreds of thousands or millions of users in jeopardy (because of weak SSL keys generated with insufficient randomness). I have no idea what the long term effects of commenting out those two lines of code are, so I wouldn't be comfortable distributing the OSE binaries I built to anyone anyhow.
On that topic - I'm not sure whether *any* binaries built of VirtualBox could legally be distributed under the GPL, anyhow - I'm worried about the fact that it depends on the DirectX SDK and Windows Driver Kit - would the terms of either of those 'poison' the binaries?
I should, I suppose, mention that it's possible that since the version of the source that I downloaded, the VBox developers may have fixed the compile issue, but the whole thing just reeks of trying to appear to be GPL, while making it practically impossible for most users (on Windows, at least) to get it working from source, starting with the fact that you can't compile it on Windows without Visual C++, and continuing on to the un-compilability of the source code version which was released at the time I tried to build the binaries ( about a month ago ).
This is not quite correct - in the FAQ they state that you can use it on work machines and still have it count as personal use (even if it is used for buisiness purposes). However, if you make an install image and roll out to 1000 users, that would count as an enterprise install. See Virtual Box FAQ, point 6. [virtualbox.org]
Overall, I think this is quite a fair license and restriction.
I've been using the non-GPL version since before Sun acquired them to run XP-only work software under Linux on an 1.5GHz Athlon, with decent performance.
The weird thing is that the boot time for XP in the virtual machine is shorter than on the real one.
What would be so odd about that? With a real XP install a cold boot has to go through A) The BIOS (about 3-4 seconds) B) The bootloader (depends) and C) The actual boot up. With a VM you only have to do C. And that isn't including any tweaks that the VM authors have done to speed up XP.
Virtual machines have their own [very fast] BIOS and bootloader. The only exception is when you run a Linux kernel from an intelligent tool like QEMU/KVM or Xen which can load a kernel from the host and inject it into the virtual machine to boot the guest.
The fact remains that real devices have warmup sequences which cannot be altogether avoided. The closest the world has come to VM-like booting is LinuxBIOS, which cuts down the device initialisation to the point that Linux can boot on top almost instantly, just like in a virtual machine.
I find this to be an excellent VM that continues to make a lot of progress. After using VMWare server, Bochs, and QEmu, this one really takes the cake on both performance and usability. Virtual machines are easy to set up using a nice graphical interface, and all of the bells and whistles require no extensive configuration (sound, mouse integration). Running a Gentoo hardened Linux on amd64? No problem. Some of the features that really put VirtualBox above the rest for me:
Sun has consistently appeared to be one of the largest corporate supporters of OSS, and their hardware is rock solid, yet they seem to get bashed every time they come up. It seems like they've been busy giving away the keys to the castle so to speak, but it never seems to be enough. What does everybody have against Sun?
Sun has consistently appeared to be one of the largest corporate supporters of OSS, and their hardware is rock solid, yet they seem to get bashed every time they come up... What does everybody have against Sun?
Personally, I appreciate Sun's OSS work. I do understand some of the sentiment though. Sun often seems to be a day late and a dollar short in their OSS ventures. They waited to release OpenSolaris under a reasonable OSS license until Linux had completely dominated that niche. Ditto with many other technologies. Even now, it is a real pain in the butt to actually get a copy of OpenSolaris and install it as a normal user. They make you install a proprietary download manager and give them a bunch of personal info. On almost all of their projects, developers not working at Sun complain about how hard it is to get changes and contributions added to those projects, because of all the red tape. Sun's OSS motto might be "we'll do OSS if we have no other option, and then we'll make it annoying". In this case they've made the binaries for this project unavailable for corporate users in a clear attempt to try to make things artificially hard so they can make money on unnecessary service contracts, instead of making it easy and concentrating on service contracts where they can provide real value (the former strategy often resulting in lesser adoption of their projects, to the detriment of said project).
I'd like to stress that I do appreciate their work. Unlike another person replying, I have no problem with their creating and profiting from both proprietary and OSS projects. They just are a big business that despite being a large OSS contributor, does not play very well with individuals or the OSS community as a whole. It leaves a lot of us personally frustrated with them when we expect them to behave like other big OSS contributors. Heck, even Apple is easier to collaborate with.
by Anonymous Coward
on Wednesday July 30 2008, @05:13PM (#24409429)
The short answer is that Sun won't get on the Linux bandwagon.
The slightly longer answer is that they are actually trying to compete with Linux. And some people will even say that Solaris is, in some ways, better than Linux. That's apostasy of the highest order for the Slashdot crowd.
A longer answer still is that most people on Slashdot are probably exposed to the worst of Sun as part of their jobs: the 10-year old behemoths. They haven't been updated in years (if ever). You can't buy parts for them, and even if you could, they're a bitch to work with because they weigh a million pounds. You spend a large part of your day just trying to keep them shuffling off this mortal coil for just a few more days. And you still often get calls about them in the middle of the night. Then you turn to your fellow admin, the guy who runs 100 shiny new Dells with RHEL5. Who has 100 times as many servers as you, but spends his entire day reading Slashdot. And you burn with hate for Solaris. It's not fair -- a 10-year old Linux box is going to be in a far worse state than a 10-year old Sun box -- but it is the way people think.
I guess the really short answer is: "A lot of reasons, none of them very good."
Sun was a proprietary vendor for quite a long time. Practically the whole reason that they take so long between announcing something is going to be open source (eg, Solaris and Java) and actually getting it into the public, is auditing the entire source tree to make sure they don't release some component licensed from some other company when they're not supposed to do that.
For example, we know the main reason why Apple went on an open-sourcing binge when OS X was released, to keep Apple relevant, but Sun never really had a down time like Apple did around the OS 9 era.
The main reason why OS X has so much open source has nothing to do with "an attempt to keep Apple relevant", it was because when NeXTStep (OS X's ancestor, why do you think most of the API still begins with NS?) was made, Unixes that were based on BSD Unix were the de-facto standard, and the Mach microkernel was considered state of the art.
There were a *lot* of Unixes that were partially open source (though this predates the open source movement) and partially proprietary at the time. OS X simply has heritage from a codebase that was state of the art Unix circa the late 80s. (Predating Linux by several years.)
I've been reading the Sun ceo's blog lately, and it seems like every post talks about open source at some point or another.
I have to agree. Jonathan Schwartz is a very brilliant individual, and his blog entries make for informative and often entertaining reads. His take on applying FOSS in the corporate world is very interesting, particularly in these times where the technological world seems to be moving away from proprietary software.
I very much believe that part of the reason Schwartz is so vehement about open sourcing Sun's offerings is partially the result of genuine goodwill. However, I also believe that much of the reason is due in no small part to his desire to a) keep Sun relevant in the news (it works to get headlines!), b) when he mentions Sun's GPL/OSI-approved software, he tends to also press the issues of maintainability, dependability, and Sun's commitment to continued support (i.e. it's open source and we can fix it if something goes wrong), and c) I think it may also be partially viral. One merely has to take a glance at the various languages (especially in the web development/scripting sphere) to understand how open sourcing the interpreter, virtual machine, or compiler tends to bolster a product's popularity. Yes, there are certainly failures in this regard, but considering Python, Perl, PHP, Ruby, LUA, and company (let's not forget the popularity of gcc when it comes to C/C++!), the only thing that surprises me is that Sun didn't open source Java sooner.
Schwartz is a good man, and I'd like to believe that while he's looking out for maintaining Sun's relevance in the years ahead as well as pushing their own product offerings and support, he's also doing good for the community as a whole. As other posters stated before, it's really a moot point getting on Sun's case; there are serious, often frightening legal implications when you open up your source--especially if you licensed parts of it from other companies. It isn't that Sun wishes to do anything evil, it's just that their hands are tied by companies that don't exactly see the world in the same light as the rest of us do (remember the fiasco regarding Java's sound libraries and the Dolby or THX issues? that's a good example).
Bravo to Sun. VirtualBox is an awesome product, and I'm glad that they've added it to their product portfolio.
"demonstrates performance visually indistinguishable from VMware"
what? i have been running vmware on my linux workstation at work for years and recently switched to virtualbox and realized that virtualbox is in orders of magnitude snappier, faster and less ressource-intensive than vmware.
just the fact that mouse support works absolutely flawless in vb is an enormous advantage over vmware. i am not even going into how much i/o wait vmware seemed to cause all the time which vb simply doesn't (yes the settings are comparable:>)
NEVER will i go back to vmware again (at least not on the desktop)
So, can I run my xbox through it? I need to be able to run simultaneously: 1) xbox halo 2) mac for screen grabs and skype 3) red hat terminals for server access 4) windows for outlook and skype
Plus, I need to be able to take screen grabs in any one of these virtual environments and save them into one or more of the others.
Bonus points if it has 'arrange by penis' for the desktop environments.
for benchmark information about virtualbox vs kvm vs vmware workstation, you might be interested in http://dipconsultants.com/press/24508-1/ [dipconsultants.com]
VirtualBox's greatest failing is that in using QEMU's I/O and networking code, they've made it a royal pain to set up bridged-mode networking on Linux hosts. You get to write two scripts, to add and remove a TAP device from a host-side bridge, and get to set up said bridge on the host yourself. Not only this, since the 2.6.18 kernel you need to run VirtualBox VMs as root (or set up sudo with/etc/sudoers not to prompt for a password and use it within your scripts), because only the superuser can manipulate the TAP/TUN devices; chmodding them writable by a particular privileged group is insufficient.
Compare to VMware, which handles all the bridging etc. by itself—much more convenient to use.
Then there are VirtualBox's "Guru meditations", obscure ERROR_MESSAGES_THAT_LOOK_LIKE_THIS and provide minimal information, often requiring perusal of the source code to figure out what's wrong. This is entirely unsuitable for end users as well as people whose time is valuable.
Finally, I tend to run a 64-bit kernel with 32-bit userspace. VirtualBox does not support this combination—it's either 32-bit kernel with 32-bit userspace or 64-bit kernel with 64-bit userspace. (VMware on the other hand does support 64-bit kernel with 32-bit userspace; its failing is that [as far as I know] there is no non-beta 64-bit userspace for VMware yet, though this will change with the release of VMware Server 2.0 and VMware Workstation 6.5.) This is only really a problem on Debian and Debian-derived distributions like Ubuntu, whose package manager (dpkg) is too incompetent to handle multiarch properly, despite work ongoing for about four years [debian.org] now, so the user has to set up a 64-bit chroot environment [virtualbox.org]. (Fedora, RHEL and CentOS get this right; rpm can handle multiarch properly, so it's only a matter of installing the appropriate libraries there.)
VMware also supports 64-bit guests on certain processors. VirtualBox doesn't support 64-bit guests at all.
So in my view, between the two, VMware still wins, open source or no open source.
InnoTek? (Score:5, Funny)
Umm, yeah. Did you get the memo?
Re:InnoTek? (Score:5, Funny)
No I'm over at Penitrode now. did the place burn down or something?
Parent
Re:InnoTek? (Score:4, Funny)
Parent
Re:InnoTek? (Score:4, Funny)
root@SanityInAnarchy:~> modprobe -r pedantry
root@SanityInAnarchy:~> modprobe humour
Parent
Re:InnoTek? (Score:4, Funny)
E: Couldn't find package human-module-humor
root@SanityInAnarchy:~# apt-get install fortunes-off
Close enough.
Parent
it was released before sun bought it (Score:4, Informative)
Re:it was released before sun bought it (Score:5, Informative)
Parent
Re:it was released before sun bought it (Score:4, Informative)
Sun just get better and better, I'm slowly finding myself using more and more of their software, and most of it is excellent.
Parent
Mentions comparible speeds to VMware... (Score:5, Interesting)
Re:Mentions comparible speeds to VMware... (Score:4, Interesting)
I had nothing but problems with it when I was testing it a couple of months ago. I couldn't get the networking to work in NAT mode, and bridging mode on a laptop ain't always the best idea. Maybe I'll give it another shot.
Parent
Re:Mentions comparible speeds to VMware... (Score:5, Informative)
I couldn't get the networking to work in NAT mode, and bridging mode on a laptop ain't always the best idea.
There was a nice bug in 1.6.0 that severely hindered networking, it has been fixed in 1.6.2 though. I only had problems with bridges and tun devices, I didn't try NAT, the bug reports had windows hosts and Linux guests, my situation had Windows and Linux guests on a Linux host. To summarize the bug: networking works perfectly until you reboot the VM, then there is no working network.
Parent
Re:Mentions comparible speeds to VMware... (Score:5, Informative)
Getting the networking system to work is a bit of a pain, but I've only had minor difficulties when using the host interface. NAT will work, but you won't be able to ping or access any resources in your own network (which is a bad thing if you have a fileserver at home and wish to access it on a VM). There are, however, a few tutorials that can help you get started with bridging your network for Windows hosts [wordpress.com] or a variety of Linux [gentoo-wiki.com] hosts [ubuntu.com].
FreeBSD is the only guest OS I've had difficulties with (even MSDOS will work, but it requires some additions to prevent it from eating up your cycles like crazy--FreeDOS plays nicely, though). I could only ever get the NAT-based networking to work and even then it would freeze whenever IO operations peaked.
Take a look at some of those articles, and you might be able to get networking up and running in VirtualBox! I have to say, for something of a FOSS offering, it's really nice.
Parent
works just fine for USB (Score:4, Informative)
Linux webcam support is problematic whether you're trying to get it on a real or a virtual machine. Has your webcam worked on any Linux physical box you've tried it on?
I'm planning to replace VMware Server with VirtualBox completely on this box. (Debian Lenny host)
Parent
Re:Mentions comparible speeds to VMware... (Score:4, Interesting)
VMWare supports multiple CPUs (2 cores visible on Guest OS) and also supports hardware accelerated 3D. Have you tried running any 3D or multithreaded apps under VMWare and VirtualBox? I find that VMWare is quite fast if you install the VMTools in the guest OS and the integration (cross VM copy/past / drag and drop, seamless mouse pointer, etc) is quite nice.
One of the main things I like about VMWare is the "Snapshot" capability which lets you create multiple "restore" points (in an easy to use visual "tree" manager) that you can instantly return to. In fact you can have a VM automatically revert to a snapshot. Does VirtualBox have any sort of advanced snapshot management?
Parent
Re:Mentions comparible speeds to VMware... (Score:4, Insightful)
I have a client running MSSQL and Exchange as virtual servers (both multithreaded apps that make good use of multiple processors/cores) and the performance was actually better with single virtual proc than assigning multiple cores to each VM.
VMWare supports processor "pinning" however and allows you to dedicate a specific proc/core to a VM which can really boost performance.
Parent
Re:Mentions comparible speeds to VMware... (Score:5, Informative)
With reservations.
You can't have snapshots of RAW disk images. it's also widely acknowledged (see the VB forums) that snapshot management is a weak point.
If you need snapshots, wait a few months/years until it works solidly
Parent
Re:Mentions comparible speeds to VMware... (Score:5, Informative)
I'm sure there was a "no" post somewhere that was informative :D
No
Parent
Re:Mentions comparible speeds to VMware... (Score:4, Funny)
Too late. It's happened.
This shows that the moderation system sucks. The post is +5 informative and yet it's clearly false.
Parent
Re:Mentions comparible speeds to VMware... (Score:5, Informative)
The documentation states otherwise [vmware.com] and you will notice the release date for VMware workstation 5.5.2 -- with 64bit Host support -- was May 2006. I have used a 64bit Host OS for VMware workstation for nearly that long.
If you are stating the free ESX Server does not support 64bit Host OS, the GSX documentation [vmware.com] from December 2005 specifically states Windows Server 2003 x64 can be used as the host OS.
Have you tried looking at VMware lately?
Parent
so what kind of VM is this (Score:5, Interesting)
Does this thing run the VM as some sort of hypervisor underneath the OS or does it piggyback the other OS's on a parent OS.
If It's a hypervisor like thing where all the OS's' are symmetric then I guess it must be getting in the way of my "normal" OS and limiting it to single core?
If it's not a hypervisor/symmetric VM and one OS is the master, Do all the OS's have full access to the hardware functions. So for example if I my mac is the master OS, and I set up a firewall set, does the windows OS have to go through the mac's firewall (and thus be protected better) or does it have direct access to the ports itself. If the latter who negotiates the conflicts when both want the CD or audio port.
Finally, are the VMs portabel from machine to machine. Or even platform to platform.
So If I create a VM on one machine, save it's state and open it on another machine, does it just run? (even the network settings?) What if the second machine was say an AMD and the first an Intel. What if the first host was a mac and the second host a linux machine?
Parent
Re:so what kind of VM is this (Score:5, Informative)
It's not a hypervisor. It's basically just an application that you run under your real OS.
The main OS treats it as an ordinary application. The primary OS firewall will effect hosted machines, for example, and as far as audio, as long as your primary OS can deal with multiple applications playing sound, then it's a non issue. Otherwise, it happens as any other conflict would. Generally, first to open the device wins.
Yes, you can move the VM images around. Part of the whole point of the VM is that it is running on the virtual hardware, and doesn't have the ability to know what the physical hardware is.
Parent
VirtualBox! (Score:4, Informative)
The best virtualization I've found for windows hosts. Works great - I run Vista Ultimate host & Ubuntu guest in seamless mode on my laptop and everything is still fast as hell!
Starcraft runs fine (Score:4, Informative)
Wow! (Score:5, Funny)
Darkhorse (Score:5, Insightful)
It looks like a viable candidate for a VM, but still a bit behind the leaders. VMWare and Parallels seem to be better choices if you can afford them, but hopefully being free as in beer and GPL will allow it to catch up rapidly and make the ongoing competition even better. If they can get 3D graphics card support running, I will be looking really hard at VirtualBox.
Re:Darkhorse (Score:4, Informative)
VMWare and Parallels seem to be better choices if you can afford them
vmware server edition is free, barring a registration via email. At least it was 3 months ago...
Parent
Re:Darkhorse (Score:5, Informative)
Parent
Binaries not Free (Score:5, Informative)
The binaries are not Free for corporate use. The source is free (GPL) but good fucking luck compiling it on a windows machine. Maybe you could compile it on a linux machine but on windows it assumes a development environment complete with every freakin' thing under the Sun (no pun intended). I gave up after two days of trying to get it to work.
Re:Binaries not Free (Score:5, Insightful)
You're already using Windows -- what's so odd about buying a license?
Parent
Contribution from non-FOSS users? (Score:5, Insightful)
It's not odd if you look at it from a contribution standpoint.
The FOSS community contributes to VirtualBox directly through help with development, testing and bug fixing on the project, as well as indirectly through their efforts on all the other FOSS projects upon which VirtualBox depends, including toolchains and mountains of utilities. Availability of source code is clearly not optional for this.
Windows binary users get a bit of a free ride on the back of all that hard work, so instead they contribute to VirtualBox by providing a bit of cash. They don't need access to the source code nor a build environment for this, and what's more, in the Windows environment it's very normal and expected to pay for your packages.
So, the VirtualBox product offering seems quite well adjusted to its two communities, and quite fair as well.
Parent
Re:Binaries not Free (Score:5, Insightful)
Parent
I like that bussiness model (Score:5, Insightful)
That bussiness model seems pretty fair to me. Release the code GPL, free binaries for non commerical use, and sell the binaries for corporate clients. They are essentially charging companies for the time and expertise it takes to compile it. And presumably it means they only have to offer support to paying corprorate customers.
A nice thing about that model is that it caps the price at the value added. Think sun is charging too much? compile it yourself and support it yourself. The value contained in the code itself, and value added to the code by unpaid GPL contributors, is not part of the price this way.
And that's a very nice way to make money off GPL. You're not cheating the contributors at all. And anyone can go into competition with sun for the compiling. So it comes down to charging for the value added by sun in compiling and servicing it.
Not quite the same as RedHat's model but highly simmilar
Parent
Re:Binaries not Free (Score:5, Interesting)
Go recursive / self-hosted build. You could always set up a VirtualBox VM with the appropriate development environment to build VirtualBox
Parent
It get's even better - the source *won't* compile (Score:5, Interesting)
I too ran into this problem where I wanted the OSE (Open Source Edition) GPL binaries on Windows. I already had Visual Studio installed, so that wasn't a big deal, but one of the requirements to build is having the MinGW g++ compiler, so now you have a situation where you need two seperate c++ compilers to compile the thing, which is kind of wierd. On top of that you need to download and install the DirectX SDK and the Windows Driver Kit, along with several open source libraries (ok, needing various library dependencies is kind of of par for the course though).
After finally getting everything downloaded and unpacked into a build tree, and getting all the command line arguments for their configure script (so it would know where to find all the libraries), the build process ran for about 1/2 hour then died with a type casting error related to the USB device driver. Now, according to the VirtualBox website, the USB wasn't even supposed to be part of the Open Source Edition (and I suspect that might be part of why I got the errors - because it was expecting it and it wasn't there).
I asked on the VirtualBox forums and developer mailing list, and after a week someone said that they got it to build by commenting out the 2 lines that generated the build error. But now I'm *very afraid*. A Debian developer who 'got rid of build errors' by commenting out 2 very critical lines of source code put hundreds of thousands or millions of users in jeopardy (because of weak SSL keys generated with insufficient randomness). I have no idea what the long term effects of commenting out those two lines of code are, so I wouldn't be comfortable distributing the OSE binaries I built to anyone anyhow.
On that topic - I'm not sure whether *any* binaries built of VirtualBox could legally be distributed under the GPL, anyhow - I'm worried about the fact that it depends on the DirectX SDK and Windows Driver Kit - would the terms of either of those 'poison' the binaries?
I should, I suppose, mention that it's possible that since the version of the source that I downloaded, the VBox developers may have fixed the compile issue, but the whole thing just reeks of trying to appear to be GPL, while making it practically impossible for most users (on Windows, at least) to get it working from source, starting with the fact that you can't compile it on Windows without Visual C++, and continuing on to the un-compilability of the source code version which was released at the time I tried to build the binaries ( about a month ago ).
Parent
Re:Binaries not Free (Score:5, Informative)
This is not quite correct - in the FAQ they state that you can use it on work machines and still have it count as personal use (even if it is used for buisiness purposes). However, if you make an install image and roll out to 1000 users, that would count as an enterprise install. See Virtual Box FAQ, point 6. [virtualbox.org]
Overall, I think this is quite a fair license and restriction.
Cheers
Parent
Works for me (Score:5, Informative)
The weird thing is that the boot time for XP in the virtual machine is shorter than on the real one.
Re:Works for me (Score:5, Insightful)
Parent
Re:Works for me (Score:5, Informative)
Virtual machines have their own [very fast] BIOS and bootloader. The only exception is when you run a Linux kernel from an intelligent tool like QEMU/KVM or Xen which can load a kernel from the host and inject it into the virtual machine to boot the guest.
The fact remains that real devices have warmup sequences which cannot be altogether avoided. The closest the world has come to VM-like booting is LinuxBIOS, which cuts down the device initialisation to the point that Linux can boot on top almost instantly, just like in a virtual machine.
Parent
A Good VM (Score:5, Informative)
I find this to be an excellent VM that continues to make a lot of progress. After using VMWare server, Bochs, and QEmu, this one really takes the cake on both performance and usability. Virtual machines are easy to set up using a nice graphical interface, and all of the bells and whistles require no extensive configuration (sound, mouse integration). Running a Gentoo hardened Linux on amd64? No problem. Some of the features that really put VirtualBox above the rest for me:
Best of all, it's FOSS.
Sun (Score:5, Interesting)
Sun has consistently appeared to be one of the largest corporate supporters of OSS, and their hardware is rock solid, yet they seem to get bashed every time they come up. It seems like they've been busy giving away the keys to the castle so to speak, but it never seems to be enough. What does everybody have against Sun?
Re:Sun (Score:5, Insightful)
Sun has consistently appeared to be one of the largest corporate supporters of OSS, and their hardware is rock solid, yet they seem to get bashed every time they come up... What does everybody have against Sun?
Personally, I appreciate Sun's OSS work. I do understand some of the sentiment though. Sun often seems to be a day late and a dollar short in their OSS ventures. They waited to release OpenSolaris under a reasonable OSS license until Linux had completely dominated that niche. Ditto with many other technologies. Even now, it is a real pain in the butt to actually get a copy of OpenSolaris and install it as a normal user. They make you install a proprietary download manager and give them a bunch of personal info. On almost all of their projects, developers not working at Sun complain about how hard it is to get changes and contributions added to those projects, because of all the red tape. Sun's OSS motto might be "we'll do OSS if we have no other option, and then we'll make it annoying". In this case they've made the binaries for this project unavailable for corporate users in a clear attempt to try to make things artificially hard so they can make money on unnecessary service contracts, instead of making it easy and concentrating on service contracts where they can provide real value (the former strategy often resulting in lesser adoption of their projects, to the detriment of said project).
I'd like to stress that I do appreciate their work. Unlike another person replying, I have no problem with their creating and profiting from both proprietary and OSS projects. They just are a big business that despite being a large OSS contributor, does not play very well with individuals or the OSS community as a whole. It leaves a lot of us personally frustrated with them when we expect them to behave like other big OSS contributors. Heck, even Apple is easier to collaborate with.
Parent
Re:Sun (Score:5, Interesting)
The short answer is that Sun won't get on the Linux bandwagon.
The slightly longer answer is that they are actually trying to compete with Linux. And some people will even say that Solaris is, in some ways, better than Linux. That's apostasy of the highest order for the Slashdot crowd.
A longer answer still is that most people on Slashdot are probably exposed to the worst of Sun as part of their jobs: the 10-year old behemoths. They haven't been updated in years (if ever). You can't buy parts for them, and even if you could, they're a bitch to work with because they weigh a million pounds. You spend a large part of your day just trying to keep them shuffling off this mortal coil for just a few more days. And you still often get calls about them in the middle of the night. Then you turn to your fellow admin, the guy who runs 100 shiny new Dells with RHEL5. Who has 100 times as many servers as you, but spends his entire day reading Slashdot. And you burn with hate for Solaris. It's not fair -- a 10-year old Linux box is going to be in a far worse state than a 10-year old Sun box -- but it is the way people think.
I guess the really short answer is: "A lot of reasons, none of them very good."
Parent
Re:Sun (Score:5, Informative)
Sun was a proprietary vendor for quite a long time. Practically the whole reason that they take so long between announcing something is going to be open source (eg, Solaris and Java) and actually getting it into the public, is auditing the entire source tree to make sure they don't release some component licensed from some other company when they're not supposed to do that.
Parent
Re:Sun (Score:4, Interesting)
For example, we know the main reason why Apple went on an open-sourcing binge when OS X was released, to keep Apple relevant, but Sun never really had a down time like Apple did around the OS 9 era.
The main reason why OS X has so much open source has nothing to do with "an attempt to keep Apple relevant", it was because when NeXTStep (OS X's ancestor, why do you think most of the API still begins with NS?) was made, Unixes that were based on BSD Unix were the de-facto standard, and the Mach microkernel was considered state of the art. There were a *lot* of Unixes that were partially open source (though this predates the open source movement) and partially proprietary at the time. OS X simply has heritage from a codebase that was state of the art Unix circa the late 80s. (Predating Linux by several years.)
Parent
Re:Sun (Score:5, Interesting)
I have to agree. Jonathan Schwartz is a very brilliant individual, and his blog entries make for informative and often entertaining reads. His take on applying FOSS in the corporate world is very interesting, particularly in these times where the technological world seems to be moving away from proprietary software.
I very much believe that part of the reason Schwartz is so vehement about open sourcing Sun's offerings is partially the result of genuine goodwill. However, I also believe that much of the reason is due in no small part to his desire to a) keep Sun relevant in the news (it works to get headlines!), b) when he mentions Sun's GPL/OSI-approved software, he tends to also press the issues of maintainability, dependability, and Sun's commitment to continued support (i.e. it's open source and we can fix it if something goes wrong), and c) I think it may also be partially viral. One merely has to take a glance at the various languages (especially in the web development/scripting sphere) to understand how open sourcing the interpreter, virtual machine, or compiler tends to bolster a product's popularity. Yes, there are certainly failures in this regard, but considering Python, Perl, PHP, Ruby, LUA, and company (let's not forget the popularity of gcc when it comes to C/C++!), the only thing that surprises me is that Sun didn't open source Java sooner.
Schwartz is a good man, and I'd like to believe that while he's looking out for maintaining Sun's relevance in the years ahead as well as pushing their own product offerings and support, he's also doing good for the community as a whole. As other posters stated before, it's really a moot point getting on Sun's case; there are serious, often frightening legal implications when you open up your source--especially if you licensed parts of it from other companies. It isn't that Sun wishes to do anything evil, it's just that their hands are tied by companies that don't exactly see the world in the same light as the rest of us do (remember the fiasco regarding Java's sound libraries and the Dolby or THX issues? that's a good example).
Bravo to Sun. VirtualBox is an awesome product, and I'm glad that they've added it to their product portfolio.
Parent
performace (Score:5, Interesting)
"demonstrates performance visually indistinguishable from VMware"
what? i have been running vmware on my linux workstation at work for years and recently switched to virtualbox and realized that virtualbox is in orders of magnitude snappier, faster and less ressource-intensive than vmware.
just the fact that mouse support works absolutely flawless in vb is an enormous advantage over vmware. i am not even going into how much i/o wait vmware seemed to cause all the time which vb simply doesn't (yes the settings are comparable:>)
NEVER will i go back to vmware again (at least not on the desktop)
straight on to boing boing (Score:5, Funny)
So, can I run my xbox through it? I need to be able to run simultaneously:
1) xbox halo
2) mac for screen grabs and skype
3) red hat terminals for server access
4) windows for outlook and skype
Plus, I need to be able to take screen grabs in any one of these virtual environments and save them into one or more of the others.
Bonus points if it has 'arrange by penis' for the desktop environments.
But will it run OS X? (Score:4, Interesting)
Does it emulate whatever Apple hardware OS X checks for, or will it still need a patched OS?
benchmark information (Score:5, Informative)
VMware still wins. (Score:4, Informative)
VirtualBox's greatest failing is that in using QEMU's I/O and networking code, they've made it a royal pain to set up bridged-mode networking on Linux hosts. You get to write two scripts, to add and remove a TAP device from a host-side bridge, and get to set up said bridge on the host yourself. Not only this, since the 2.6.18 kernel you need to run VirtualBox VMs as root (or set up sudo with /etc/sudoers not to prompt for a password and use it within your scripts), because only the superuser can manipulate the TAP/TUN devices; chmodding them writable by a particular privileged group is insufficient.
Compare to VMware, which handles all the bridging etc. by itself—much more convenient to use.
Then there are VirtualBox's "Guru meditations", obscure ERROR_MESSAGES_THAT_LOOK_LIKE_THIS and provide minimal information, often requiring perusal of the source code to figure out what's wrong. This is entirely unsuitable for end users as well as people whose time is valuable.
Finally, I tend to run a 64-bit kernel with 32-bit userspace. VirtualBox does not support this combination—it's either 32-bit kernel with 32-bit userspace or 64-bit kernel with 64-bit userspace. (VMware on the other hand does support 64-bit kernel with 32-bit userspace; its failing is that [as far as I know] there is no non-beta 64-bit userspace for VMware yet, though this will change with the release of VMware Server 2.0 and VMware Workstation 6.5.) This is only really a problem on Debian and Debian-derived distributions like Ubuntu, whose package manager (dpkg) is too incompetent to handle multiarch properly, despite work ongoing for about four years [debian.org] now, so the user has to set up a 64-bit chroot environment [virtualbox.org]. (Fedora, RHEL and CentOS get this right; rpm can handle multiarch properly, so it's only a matter of installing the appropriate libraries there.)
VMware also supports 64-bit guests on certain processors. VirtualBox doesn't support 64-bit guests at all.
So in my view, between the two, VMware still wins, open source or no open source.