Slashdot Mirror
VOS Patents on Virtualizing OSs?
Erik Poupaert writes "I've been following VMWare and plex86 (used to be freemware) for a while now, because I think that virtualization may be a solution to quite a number of problems. It basically allows you to run several operating systems concurrently. IBM uses this approach too, to run a large number of Linux instances on their mainframes. While you can leave the task of managing devices and device drivers to the host operating system, guest operating systems can supply you with the ideosyncracies of their particular versions to run the applications that expect them.
Since VMWare is not free, as in free speech, I thought that plex86 would become the lead open-source project in the field.
Now there seems to be a new player, that I never heard of before, called VOS, who claims the whole field to themselves, and have filed patents to obtain a monopoly on the entire discpline. Have they got any chance in succeeding? Or do you think that the patent office would not grant such patent?"
- There is *no* trial download available, despite the "press release" saying that the product was released a *year* ago. I'd think
/. would find it before that.
- If you look at the site structure, everything is flat-file, i.e. it's all page_XX.htm. Highly unprofessional site.
- The "buy now" page is insecure, so I wouldn't try it.
- They say that USA Today, etc. have commented on Flash VOS; I just tried a google search and nothing showed up on the first page.
If I'm wrong, someone please post a clarification. The only thing is that the domain seems legit: a whois lookup shows that it was registered in 1998...--bdj
From their Homepage:
This looks like a BIOS-Bootmanager, it just activates the partition the OS you're interested in resides on, then let's you boot from that partition. Magic indeed.
Aah, a BIOS based bootmanager with APM - The state of your work is saved to disk, then you reboot.
Okay, that looks like a VMware done in hardware. Each OS is given a slice of the available RAM in which it can reside and run. You can switch between different Virtual-Machines. Okay, ther is no emulation layer in here, as all OSes are running on the actual hardware. When switching OSes, the OS you're switching from has to release all Hardware - that seems to be interesting. But that is "Phase III", so I wouldn't hold my breath.
Can't wait for that one - does anyone need a nice bridge? I own one I could sell :)
Ralph
Hasn't anyone been paying attention to the patent wars that have been going on? The patent will pass, do not doubt it. The US Patent Office is too stupid to catch it. "Oohh a way to make a computer pretend it's a different kind of computer. No one's ever done THAT before!"
The patent will remain until it is challenged in court and defeated, and it's almost impossible to overthrow a patent in court. The 800 lb gorillas can't start anything unless VOS tries to enforce the patent against them.
The end result? VOS gets the perfect weapon with which to kill off smaller competitors as long as they leave IBM and the like alone. The IBM bean counters won't authorize a legal battle over the patent just to keep a few Open Source projects alive. To them, it's not worth the expense. Plex86 is either killed off or developed only by people who live in countries that laugh at US patent laws. VMWare will probably be safe, but if VOS develops an attitude and gets some venture capital behind their legal department, VMWare will be toast.
Matthew Miller,
"Live Free or Die." Don't like it? Then keep out of the USA
Basically "unsafe" code in this context is any code that tries to access hardware (should be possible to trap with the MMU), and code that uses some x86 trickery to find out which "ring" they run in, which is impossible to properly virtualize, and has to be emulated.
Since the "unsafe" code cases are fairly uncommon, this delivers a significant performance boost over emulating all instructions...
How about someone starts a competition to come up with something the USPO wouldn't allow. The funnier the better. If the word got out to the mainstream media as a filler item, it might even make a useful point to the masses and their masters on The Hill.
Of course, we'd have to have a prize worth noting, so it would need some rich sponsor with an interest in the topic (Hello, RedHat; VA Linux?).
Rational debate isn't working with the PO, so maybe it's time to make fun of them in public.
TWW
"Encyclopedia" is to "Wikipedia" what "Library" is to "Some people at a bus stop"
When faced with a set of patents like this, you have to hope that there is an 800lb gorilla out there who might take exception. Given that IBM's S/390 machines are just about as good as you can get in Virtual Machines either through VMs or LPIs, anything which seriously treads on the toes of IBMs patents in this field is likely to get short shrift in the legal arena. That said, this is already a heavily patented area - a quick search of the patent database pulls up 245 patents on this issue. Which is pretty scary given that these patents, as with so many patent applications, aim to be as broad as possible in their presentation.
So much for propelling innovation forward...
Cheers,
Toby Haynes
Anything I post is strictly my own thoughts and doesn't necessarily have anything to do with the opinions of IBM.
I don't believe these patents in any way cover "the whole discipline"... and if they do there is plenty of prior art to see the patents off.
IBM's VM (AKA OS/390) has been doing virtual machines for longer than most Slashdot readers have had a computer. Emulators (and I was playing with XZX as long ago as 1995 -- it probably was around a while before then) are essentially no different.
All VMware did that was clever was to virtualise an x86. Virtualise (as opposed to emulate) because it does not need to emulate CPU operations, merely pass them on to to the real CPU. An S/390 is designed to facilitate this. For reasons I don't understand personally, apparently a PC isn't particularly friendly to this approach, so VMware did a great job in pulling it off.
I really like VMware, I use it every day (to run Lotus Notes. Bah.) - a Free equivalent would be lovely.
--
The notion of a virtual OS layered on another OS is not new. Virtualizing RSX11M was proposed in public at the Miami DECUS meeting by Ralph Stamerjohn; I believe this was in 1980 but it could have been as late as 1982. I published the MSX-11 OS which had a mode in which it ran as a virtual OS under RSX11M. That was written in 1979. Full source was published. Virtual disks were an implementation of restrictive partitioning. Mine for RSX11D and Ralph's for RSX11M were published initially in the late 1970s. The concept of virtualizing the device on something completely invented existed in publications by the mid 1970s (gave me the idea about the RSX11D virtual disk). The patent office may never have heard of DECUS, let alone looked at its publications, but they were and are open to the public. (DECUS membership has been free of charge also). This is different from a virtual machine. Incidentally, the RT11 under RSX implementations first surfaced, doing a complete OS under another OS on the same machine, by 1979. I had a copy of RT11 V2C running under RSX11D in 1979 and there had been earlier ones. (Had to slow down or stop the clock though; the host machine and os couldn't keep up.) The virtual RSX11M was intended however to virtualize at system service level optionally, not just at driver level. While it was never wholly implemented (we had day jobs) it was discussed in detail and some bits like virtual debuggers did get implemented. I am sure there were many others.
Unfortunately I cannot see the patents themselves. But there is so much prior art in this area that even if the USPTO will do its usual thing they do not stand a chance in court.
IBM, Intel, Miscrosoft and Digital(now compaq) have enough prior art, IP and even patents as well as financial resources to simply roast them and eat at their leasure.
IMHO: They have taken laisurely pace towards Vmware and Plex86 because they have not attempted to file IP and bite them. This is different story. The patents have been filed. The IP war tomahawks have been dug out.
Plex86 and Vmware should simply wait until the legal ICBMs will hit on target and the legal fallout has finished.
That is if any of the patents that make any impact will actually succeed. As of now patent search return not found on all of them.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
At least as far back as about 1991 or thereabouts, a company called Insignia Solutions made a Macintosh product called Soft PC. It was a software emulator that ran Windows on a Mac, albeit slowly.
:-) ]
It pre-dates the PowerMacintosh which appeared in 1994. (That is, Soft PC ran on the Motorola 68000 processor.) When the PowerMac version came out, it was named SoftWindows.
IIRC, SoftPC goes back at least a couple if not more years than the PowerMac (1994).
Later Connectix came out with Virtual PC, which IMHO is a much more useful product (see below). Whereas SoftPC/SoftWindows take a lot of shortcuts and install specialized Windows drivers to enhance performance, Connectix's Virtual PC tries to truly emulate the PC hardware -- upon which you can install any OS. Virtual PC is more like VMWare, while Soft PC is slightly closer to WINE. I would date Virtual PC to be at least 5 years old.
There was another emulator product for Mac about the 1995 timeframe called Blue Light. I've never seen it, so I can't say much. But all this taken together should provide some prior art.
I've successfully installed SuSE 5.2, 6.x, Win NT 4 Server Enterprise Edition, Win NT 4 Workstation, Win 95 osr2, Win 98, 98SE, Win 2000 on Virtual PC. Haven't had time to try WinME.
[Why? Because it is the most wonderful testing environment you can imagine. I can fsck with the registry or other windows internals with total impunity. I can copy a previously backed up Drive C image from a CDROM in 4 minutes and have a totally virgin install of the OS back again. Wonderful QA tool. QA dept. can have a disk wallet full of CD's containing Virtual PC drive images with virgin installs of all your un-favorite OS's with various permutations of service packs, etc., all properly licensed. What OS do you want to fsck with today? Plus, I can install some screwball game, and do a before and after comparison of the entire hard drive image (using Mac tools outside the PC environment, to avoid Hiesenbug-type problems). I've even got drive snapshots of NT inbetween all the zillion reboots you have to perform during it's installation. Now if only Connectix would release a debug version that would allow me to stop the emulated processor in between instructions -- this would be the equivalent of hardware debugging from the perspective of the Guest OS. But I digress.
I'll see your senator, and I'll raise you two judges.
A link: search on Alta-Vista turns up Pearl 9 Design which lists Flash VOS as a client. Pearl 9 Design has a mission statement that states, "Endeavor to bear standards beyond the ordinary." And looking at their site and the FlashVOS site, their concept of ordinary must be really low.
Secondly, the other clients Pearl 9 List, don't exist (or at least the links don't work). Looking at who FlashVOS lists as business partners and the list on Pearl 9 clients...I see a large amount of overlap.
Seeing that this looks like a big ol' hoax, and the fact that they are taking credit card orders, I suspect something malicious. From the broken links, the lack of anything to download, the lack of a user guide, the lack of screenshots, one can oly conclude that this is a pure vaporware site, or something criminal. I've written to the President of Flash VOS to have him contact Slashdot to verify his company's product (which is selling for just $30!).
This is not the way to build a lasting empire.
However, here's the list of the patents they claim to have filed. So, if they are overbroad, they may be caught by the patent offic and not be granted [0].
The flash based part here I think is critical. Rembeber that patents are very specific. A flip down mouthpeice on a phone is not the same as a flip up ear piece. Thus, this seems to restrict it to using flash only. Precisly what innovation it has (other than standard viruliasation and journalling) I'm not certain of, but I don't belive that this will be a problem.
Um, lost on the techincal bits of virtualisation on this one. It sounds fairly over broad.
Not certain what they mean by restrictive partioning, as apposed to 'ordinary' partioning. Google is confused on it too.
These say they are OS specific. Shouldn't be a problem.
[0] Yeah, right.