Slashdot Mirror
Will Munich's Linux Desktops Be Running Windows?
An anonymous reader writes "Remember that story about the city of Munich choosing Linux to power 14,000 desktop computers? One aspect of this story that most people don't know about is that up to 80 percent of those Linux desktops will be equipped with VMWare, a virtual machine emulator, under which they will run Windows and Windows applications. That's right, folks: The majority of those 'Linux desktops' will be used to run ... Windows." This Gartner report from early June seems to be the one mentioned in the article, though I'm not sure exactly where Thurrott gets the 80% figure.
interesting if it's true, although the article does specify a redmond source.
vmware is a great product for developing using a virtual network, and I'm sure they must have done their sums (they are Germans, after all)
My guess is that they must have a lot of legacy apps that they can't afford to update now, and that in the future they will use linux clients, but that they can start moving everyone to linux now to make the later switch easy.
Humorous signatures are over-rated.
By switching to Linux for operating system, Munich avoids depending on Microsoft for technical support and product upgrades. While using VMWare allows continuous usage of existing programs and possible future migration to Linux. This decision allows CHOICE and FLEXIBILITY, which are what I believe Munich wants in the first place.
It often makes sense to maintain backward compatibility for a stage or two. If they were just going to run Windows apps they wouldn't need Windows at all. So clearly, if the story is even true, what it means is just that they have some 'legacy' apps they can't immediately replace. No big deal. Run VMware (or Win4Lin, or Wine, depending on the specifics) use Windows and the software for it that they've already payed for, but it gets them off the upgrade treadmill, and looking for *nix-native programs to replace the legacy crap.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Friends don't let friends enable ecmascript.
Good grief. A little common sense would dictate that even if they are using VMWare, its not something to generate a whole lot of hulaballoo about. Everyone needs a transition period.
Munich has over 100 Windows-only custom apps! They can't be expected just to switch to Linux and loose those overnight.
Even *I* dual-booted Windows next to Linux for 2 years, and I'm a geek. Its to be expected, and isn't "Betrayal" or a "Loss" to Linux... it is the necessary progression.
I think it really depends on what they're doing with the Windows VM's. If they're booting up their Linux boxes, firing up VMware and spending the duration of the day in Windows, then yeah, that's pointless.
But if they're only using Windows/MSOffice under VMware to aid in the transition to non-MS software, and using Linux everything else, this could be a huge opportunity to introduce Linux on a large scale without pissing people off with OpenOffice incompatibilities, etc... IMO, this is a good thing.
While in an ideal world, the entire setup would instantly switch to a purely linux solution, the real world doesn't work that way. You'll always have people who are more comfortable with certain apps, and you'll have applications that just plain aren't available for Linux just yet.
By running the workstation on Linux, and by locking Windows into it's own little virtual machine jail, this accomplishes several important things:
1. Gets Linux on the desktop
2. Handles any/all remote display sessions, eliminating cost of a windows X-desktop solution.
3. Handles remote NFS mounts to *nix servers, eliminating *that* cost.
4. Allows users to continue with most of their work when the Windows VM bluescreens - reboot the VM and keep going.
As long as 3 years ago, a Linux desktop with VMWare running Windows was a viable solution from a cost perspective, and with the reliability and other improvements in VMWare, it's an excellent solution in a hybrid environment.
While many of us would prefer to be in a Pure Open Environment, the reality of the world is that this doesn't exist in many places just yet. Moves such as this will go a long way torwards getting us there, and in the meantime, there are a bunch of desktops which will be far more stable and usable than if they were running only Windows. I've been running this way for around 3 years now, and it's a perfectly viable solution. It gets Linux the exposure it needs, and the quality of the product will do the rest.
Look,
Bad news would be a total reversal and going with a full MS solution. This is an intelligent, well thought out migration. As nice as it would be to have all Linux based apps at the changeover, it's just not going to happen. As nice as it would be for all systems to be rolled out with Linux and everyone be perfectly happy with their systems and no retraining required, it's not going to happen. This is an excellent migration strategy. Start porting your critical apps over to Linux over a period of time (months/years), keep your existing legacy apps running on Windows in a sandbox (ie. VMWare) so that the typical crashes don't bring down the whole system (just restart the virtual machine), and, best of all, keep everyone happy and content. This shows that IT was thinking of the users throughout the migration and have their best interests at heart.
To everyone blasting this as a betrayal or a PR stunt, work in the real world and do this for a living. Wholesale switches from one platform to another is not an easy thing from the IT side of things or (especially) the user side of things. This is warranted, justified, and just plain right.
These guys don't need yelling and screaming, they need a pat on the back for thinking this through thoroughly enough to see that not all apps can be migrated over yet and people still need to work, otherwise, they could have 14,000 new bright and shiny systems with only a handful of people actively doing their jobs while all of the others sit and twiddle their thumbs for most of the day (besides to send emails, type up some documents, and surf).
I love linux (as I state all of the time), but a migration of that size (or any size) has to be thought through from the perspective of not only IT and monetary considerations, but from the user standpoint as well.
CliffH
sigs are like a box of chocolates, they all suck remove the underscores to email me
I'm a Mac user myself. But Virtual PC has it's place. Alot of my job entails Visual C++ 6 programming, so being able to run that is nice.
I don't feel I'm betraying Mac because of it. Besides, how can you betray an inanimate object? There is a good article about fanatacism over computers. In reality they're just tools to help us accomplish an end. Whether it be saving the world, or getting that file off my digital camera and sending it to Aunt Sally.
Some people get attached to them, fight wars over them, etc... but in the end there are different computers, many capable of doing a job. Windows, Linux, BSD, OS X, IRIX...whatever. All are pretty decent.
If running Windows on Linux lets them do whatever they need to do, then who are we to criticize them?
Exactly. When transitionign to a new platform, any userbase is going to need to figure out how to transition their data over, find or create replacements for Windows apps, etc.
It's not like they're running Linux *just* to run Windows. I'm sure some of their apps will come from the Linux side, and some from the Windows side until their Windows apps are slowly phased out.
After almost 10 years of running Linux, I *still* have a small Windows XP partition for doing stuff like games or certain Windows-only apps. I spend the *majority* of my time in Linux, sure, this is still a very Windows-centric world.
My journal has hot
this start demonstrates already how serious a reader can take you: the city in question is Munich, not Frankfurt
They simply didn't want an American winning,
so IBM is not an american company? could you please let us know where their headquarters are?
and were willing to do anything, even pay more, to accomplish that goal.
i suspect most of the extra cost is training for the new environment, a one-time cost. the main reason microsoft's offer ended up being so much cheaper is that they offered to pay the extra training out of their slush fund, something that they most likely would not offer again on the next upgrade. with linux additional training on future upgrades should be considerably less.
and do you really think that the only reason MS can lose is anti-americanism?
and the only way to run Office is to use VMWare.
I use Microsoft Office both on my home PC and my work laptop (when necessary, which is, unfortunately, often). They both run Linux. I use CrossoverOffice, which is a commercial version of Wine.
So, there are ways other than vmWare, FYI.
JWall: GUI client for IPTables
Ok, in reading most of the 3+ comments here, it looks like most slashdot readers here, at least the ones with a bit of foss experience, know that having vmware on a fat client migration from windows to gnu/linux in a business setting is a requirement, and there is no such thing as immersion or shock implementations of FOSS in a large business or government, unless you are trying to guarantee failure.
What everyone should be questioning, however, is why doesn't Gartner know this? Or do they?
If you have been following the juggernaut of gnu/linux, you are aware that gartner is slanting every report they can in favor of microsoft. Quite simply, microsoft pays many of their studies. What would happen to gartner funding if they started coming out with stories that said that it is cost effective to switch to gnu/linux?
Do you think that microsoft would continue throwing the same amount of money gartner's way? Especially when execs and government workers, the majority of which don't know their asshole from a donut hole, rely on gartner's reports for purchasing and implementation decisions?
microsoft has billions to grease the wheels. Should gartner piss off microsoft?
microsoft was so desperate to not lose the munich deal that butterball ballmer cut short his ski trip to try and salvage the deal. And when they tried once again at the last second to underbid the foss deal (which they were already underbid), one of the munich officials considered it "insulting".
So now, microsoft has to do everything it can to spin this loss as best as possible. Claiming that vmware/windows will run on 80% of the desktops smacks of tactics used in the OS/2 wars, similar to the current tactics of hitting Newsforge and Slashdot boards with "Linux just isn't ready for the desktop yet". That's right. That phrase keeps popping up. When you see that phrase, review what went on during the OS/2 wars. If you are too young to remember, google the phrase.
microsoft will be spinning the munich loss for many years to come. This 80% figure, on munich desktops, is exactly this: spin. It won't even be implemented by 1Q04, and we already have spin that says that in munich, an IBM/Suse/linux win, microsoft loss, 80% of the desktops will be windows desktops because they have vmware installed.
What can Gartner do to make their reports more balanced? How about counting down time/reboots in their tco studies? Or counting how many servers a windows admin normally maintains, against how many servers a unix/linux admin maintains, in their tco studies? Or counting actual number of patches of actual number of installed and running services in comparable systems, instead of counting every patch of every application included in the major distros, even though microsoft has no functional equivalents, or is not running them on a particular installation. Does gartner include patches for microsoft office when comparing file servers? So why throw in patches for applications for gnu/linux distros that would never be included on a file server, as part of the total count?
In short, do you believe that gartner will do this? Have they done this in the past? Have you taken a look at their study that found that gnu/linux was cheaper only for web servers, and microsoft applications were cheaper for everything else on a tco basis? Have you taken a look at the criticism of this study? It's been covered here on slashdot previously.
The gartner tco study counted linux/unix implementations as costing more partially because linux/unix admins command a higher wage/salary. And higher technical skills. But they assumed that windows and linux/unix admins also administered the same number of servers. Anyone with extensive experience on both sides of the fence believe this?
What about hardware? No where in the gartner study did it mention or take into account that hardware requirements for gnu/linux is less than windows. No where.
Also, conveniently, the leng
They can take those licenses running on VM's and continue to use them for as long as they want to regardless of hardware issues.
They can limit the interactive nature of those images as well. Strip them until they only perform the tasks required. No browser, etc...
Admin can be easy if they want. They can deploy the images on a central server or on each machine.
I think the admins will be amazed. Once things are working the way they can, the setup will have clear advantages. If a user hoses things up, copy standard image, restart VM, done! --This is easy now and will remain so even as the rest of the environment changes.
Because they removed the win32 apps and software from the hardware environment, they can and will be able to run what they have now for as long as they want.
Blogging because I can...
If you read the Gartner report you will see that it doesn't refer to any sources at all.
At one point it says "Munich hasn't yet disclosed the business case behind its decision." and at another "The business case assumes that many applications will not migrate to Linux".
Note that the source of the business case is carefully omitted and it doesn't refer to Munich's business case as that hasn't been released. I read it at referring to Gartner's own ASSUMED business case.
The article is just another imaginary pro-MS Gartner report. Move along folks...
GNU/Linux desktops with VMWare for virtualized access to other systems -- be they legacy MS Windows, GNU/Linux, or other, makes tons of sense.
First, you've kicked the monkey off your back, er, desktop. The principle system is Linux, and the end-user application space is more than adequate for general business computing. Depending on the services SuSE and IBM can offer, the flexibility and management of this solution will be worlds above what a Microsoft environment could offer. And running VMWare on GNU/Linux to serve legacy MS Windows makes far more sense than hosting it the other way around given the stability, configurability, and performance of GNU/Linux.
Second, VMWare is a great product -- words I'm not prone to utter about proprietary software in general. The high-performance general system virtualization niche is one that VMWare has fully locked up. I've used the product since first public betas in 1999, and none of the other alternatives I've tried -- Bochs, Plex86, UML, WINE, or dosemu -- hold a candle to it for ease of configuration, versatility, stability, or performance.
Third, VMWare provides an awesome way to manage a large number of desktops. Within the virtual machine, the hardware configuration is identical for all systems. The only differences are processor speed, memory, and availability of specific peripherals which don't exist on the host system. However, all devices -- disks, network cards, sound, monitor, etc. -- are the same for any VMWare virtual machine. Any system-specific drivers and related configuration is a non-issue.
Fourth, VMWare allows access to multiple configurations, which may be accessed simultaneously. In part this is a box Microsoft has launched itself into with gusto. A trivial example is browsers: one side-effect of the tying, er, integration of MSIE with the legacy MS Windows product is that it's not possible to run multiple variants of an MSIE browser since version 6.mumble. With VMWare, it's possible to run different configurations of, say, Win2K and MSIE. It's also possible to run different legacy MS Windows OSs entirely: Win3.1, 95, NT, 2K, XP, and their variants. Simultaneously (a respectable system should be able to support 3-4 concurrent virtual machines if necessary). Support desks worldwide already find this invaluable. It's likely that Munich's aquired a motley mix of applications which run in a number of specific environments -- VMWare is likely one of the best ways to make these conveniently available to workers.
Fifth, the virtual system images themselves are nothing more than file snapshots. These can be stored and served centrally (again reducing maintenance issues), and eliminating again the overhead of creating and installing thousands of systems -- rather, a few standard file images are served centrally. User-specific files can be served over the network from your GNU/Linux Samba server.
Sixth, VMWare's rollback and checkpointing means that for a given image, it's possible to run a system either with no commits (all changes to the running VMWare image are lost on exit, great for highly specific tasked workstations), or can be committed or discarded as an option, on system exit (useful for development). In either case. backing up the image file prior to use allows for recovery later.
Seventh, once you've kicked that Redmond jones, "deinstalling" the now-worthless virtual system is a snap.
I'd say Munich's going to have an excellent, flexible, configurable, stable, and useful system.
What part of "gestalt" don't you understand?