Submitted by “Intel and VMware announced today that Intel Capital is taking a USD 218.5 million stake in virtualization company VMWare. Intel will purchase 9.5 million Class A shares at USD 23 per share, which, at the completion of VMware’s forthcoming IPO, will give Intel about a 2.5 percent stake in the company. Because VMware’s stock is split between Class A shares, which have less voting power, and Class B shares, Intel won’t control that many votes in the company, but they will get a board seat.”
Looks like MS is out to take over another ‘partner’s’ market. From the article –
Although I can understand MS wanting to get on board the virtualisation gravy train, mainly due to the fact that you have several virtualisation technologies out of the box on any Linux distro (not to mention built into the kernel), trying to kill off a former ‘partner’ by offering a free, Enterprise level alternative integrated into it’s consumer desktop offering just seems like more abuse of it’s virtual monopoly.
Hopefully we can have several open source implementations to offset MS if they’re offering turns out to be as bad as VirtualPC currently is. ;-P
Edit: Added some bits for clarity.
Edited 2007-07-09 22:50 UTC
The article seems to be forgetting VMware ESX, I don’t think Microsoft will be able to compete for the same level that ESX provides for a while yet.
ESX runs VMware’s own kernel and hypervisor so instead of having virtualization logic *added* to the kernel, the kernel is designed for virtualization, leading to better performance.
Microsoft may be able to convince some shops to switch to their own built in virtualization tool for some testing but I wouldn’t expect them to start taking over the market. Also you have to consider what operating systems will be supported for running in the virtualized environment, if non-Windows guests couldn’t perform well then that could be a show stopper for some.
When it comes to virtualization, the money is not in the host runtime – both MS and VMWare give it away for free – but in the managability software that surrounds it (there are strategic aspects too). Microsoft makes the System Center product line, which has virtualization concepts built deep into it and correspondingly they offer a runtime.
Microsoft definitely desires that Windows be your virtualization host (for any and all workloads), and will build good guest-level support for alternative OSes to support that goal. Their arrangement with Xen speaks to that.
Edited 2007-07-09 23:24
Although I can understand MS wanting to get on board the virtualisation gravy train, mainly due to the fact that you have several virtualisation technologies out of the box on any Linux distro (not to mention built into the kernel), trying to kill off a former ‘partner’ by offering a free, Enterprise level alternative integrated into it’s consumer desktop offering just seems like more abuse of it’s virtual monopoly.
This time around it is going to be hard to cry monopoly like before with Internet Explorer being deeply integrated into the OS since Linux has virtualization right in it’s kernel too.
Virtualization makes sense in the kernel, this is something that Microsoft should do.
I can’t be believe that VMware is worth $8.74 billions USD! This must drive the author of QEMU crazy
Yeah, sounds like a huge amount of money for just…VMWare? I didn’t know so many people used VMWare. But it must be generating tons of bucks to be worth this price tag.
Regarding KEMU, the problem is that it doesn’t generate much money, and probably doesn’t have many customers. This is what’s important when you sell your company.
Regarding KEMU, the problem is that it doesn’t generate much money, and probably doesn’t have many customers.
Actually, it does. Both KVM and Xen HVM rely on qemu bits. Maybe it doesn’t bring in much cashflow for the qemu project, but it certainly does for Red Hat, Novell, XenSource, and others who provide services in the Xen ecosystem.
(Not to mention smaller players like Win4Lin (mostly qemu-based) and VirtualBox (partly qemu-based))
As usual, the money around open source projects is not in selling code, but in selling services/support.
Edited 2007-07-10 11:06
Only $8.74 billion USD? Sounds like pocket change… 😉
“This must drive the author of QEMU crazy
”
1) QEMU is not practical for average joe (too slow for anything but testing, much less server consolidation [which is where the money is])
2) If you want to make money, don’t give away code for free
The QEMU Accelerator (aka KQEMU) is now available under the GPL. Read about it here:
http://fabrice.bellard.free.fr/qemu/kqemu-tech.html
It’s very cool stuff. QEMU is used in a variety of commercial products, and it’s a critical part of the Linux KVM architecture.
All of the virtualization vendors are scrambling to give their software away to as many users as possible in hopes of gaining a foothold in this important market. It’s more important to build marketshare than to drive licensing revenue.
It seems like a good bet for Intel as they can now create some cool combination for hardware/software virtualization system and have a better control for the market as a whole.
I hope this does not hinder Intel contribution to KVM, Xen, and other OSS virt. products. I’d rather have seen such capital used to bootstrap XenSource.
Go opensource and freeware instead , ohyeah and dump software patents they no good
What I’d like to see is VMWare for Solaris – that would make migration a whole lot easier; QEMU unfortunately doesn’t support USB devices on Solaris
You mean you don’t use Containers/Zones? And the [possibly] sad truth is that most migration is from Solaris to Linux, instead of the other way around, so virtualised Solaris is more important that virtualisation on Solaris.
And why would you want to use USB on server-class kit like Sun boxes?
If you mean Solaris as a desktop OS (yuk!) then I’m sure it wouldn’t take OpenSolaris long to get USB drivers for VMWare working.
VMWare support has come a long way. I remember hacking together X11 and NIC drivers for Solaris 8 on VMWare Workstation 3, now with Solaris 10 on VMWare Server, everything pretty much works out of the box.
As far as Windows built-in VM, well if its based on VirtualPC or VirtualServer, I don’t think VMWare will be losing any customers any time soon!
For that matter, KVM/Xen are still a long way off VMWare Server on Linux, even on Fedora7, which is pretty much built for virtualisation, its pretty crappy.
The nice thing about QEMU was that it emulates non-x86, but that’s the bit that all the forks like VB/KVM/Xen leave out. Imagine virtualised Power or Cell or ARM under VMWare, it would be lovely for developing for PS3 or OpenWRT.
That’s something I would like to see too. But since Solaris already has Zones and Branded Linux Zones there is less of an incentive for VMware to port it. And when Solaris includes Xen by default in the Solaris 11 timeframe so you can run Windows as well (at least on VT/SVM processors), VMware will be shut from another operating system.
BTW the Microsoft Viridian technology is nothing else but an adaptation of Xen to the Windows platform they had to resort to because Virtual PC/Server have horrible performance and features (not that I have ever tried them, because I run Slackware, Solaris and BSD on hosts and Windows 2000/XP occasionally in VMware guests).
I need VMWare for Windows support – now if there was a Windows brandz zone, now that would be awesome 🙂 but the sad thing, it won’t happen anytime soon 🙁
I need it primarily so that I can load music onto my MiniDisc.
I have Slackware running on all my main systems now and I have VMware installed so I can run Windows. It’s also very Unix-like in terms of stability, reliability, performance and simplicity, so it shouldn’t be very difficult to switch between Slackware, Solaris and the BSDs. To me it all looks and feels the same.
I understand this might not be your preferred choice so the next best thing would be to fix QEMU so USB works in virtual machines on Solaris. I am subscribed to the QEMU developers’ mailing list and I use it mainly to emulate other architectures to run various ports of Slackware (including my own to MIPS Loongson) and Debian.
I intended to install Solaris on my server (former desktop) because of zones and ZFS but I’ll probably refrain from doing so because I’m looking forward to migrate from x86 to Loongson 3 next year.
I figured it’s not worth the hassle if I had to migrate from ext3/xfs to ZFS and back again in a year’s time. If I or someone else could port Solaris to MIPS I would consider it again.
I can try and see what’s needed to get USB support working in QEMU guests on Solaris hosts :-). First I’ll have to download and install the latest Nevada build though.
It will be interesting to see how the MIPS processor goes – if it is fully open and documented, it’ll be a great platform to base operating systems on.
It would be interesting to see it in a laptop, especially considering the 3-8watt at 1Ghz which documentation claims; IIRC thats 5-10 watts less than the UlraSPARC IIe running at half the speed.
It is one of the most open platforms in existence. That’s why there is such a broad ecosystem of vendors for all kinds of MIPS processors not to mention that licenses are a lot cheaper than for example ARM. The only architecture that is more open is SPARC.
Many patents on the architecture are no longer valid so newer versions of the MIPS ISA can be targeted. And since ST Microelectronics has taken a MIPS64 license the upcoming Loongson 2F will be fully MIPS64R2 compliant.
These claims are right. I have received a Fu Long box from Lemote and that’s what I used to port Slackware. It is always powered on and doesn’t run hot although there is a fan that makes (little) noise all the time because of non-existent power management in the kernel at the moment.
The components on the inside are in fact the same as those in a notebook. You could see it as the equivalent of the Mac Mini G4 vs the Powerbook. The released systems are the first ones available and newer ones based on Loongson 2F, DDR2 and PCI-X will come later this year.
The 200 notebooks that have been created are only available for their internal use and development. I expect their successors to be available at the end of the year though.
If you look at historical performance statistics it was always so that Digital Alpha, HP PA-RISC and SGI MIPS came out on top with the rest following and SPARC mostly somewhere at the back of the list, so I’m not the least surprised at Loongson’s and other MIPS implementations’ performance.
I find it strange that now that more and more people are migrating to less power-hungry architectures Apple chose to go the other way around and not always to the enjoyment of their notebook customers particularly with respect to the heat the Core2Duo laptops generate. I would have taken a multicore G4 or Pwrficient over Core2Duo or Turion X2 any day of the week.
Edited 2007-07-11 09:18 UTC
I’m hoping that the voice of Intel will help to keep VMWare from making any more Windows only decisions like they did with their VirtualCenter product, and perhaps see the light on that choice. Linux is such a dominant operating system in high performance clusters and yet what did VMWare decide to use for their clustering of ESX servers? A .Net client and a Windows management server. What nonsense.
In recent years Intel has come to see that Windows only is shortsighted and hopefully they can pass that message along.
If VMware wants to they will, if not then we’ll have to look forward to other technologies such as Xen en KVM. I agree that a company making such boneheaded decisions in these times is not very wise indeed. They could have used cross-platform technologies.
But I don’t care about VMware the company. I just want a product that does what I want it to. If VMware is the company that delivers that functionality, that’s fine. If not, then I’ll have to say goodbye to VMware. I’m mostly interested in KVM on my Linux machines these days anyway.