Robert Pogson

One man, closing all the windows.

Munich Migration From The User’s Viewpoint

  • Sep 07 / 2011
  • 64

Munich Migration From The User’s Viewpoint

Spiegel has a piece about a department manager’s view of the migration. The guy is 60+ and worked 20 years with computers before migrating to GNU/Linux. Some details:

  • The major problem he encountered were some duplicated records in a database. These were eventually fixed.
  • 22 CMS (Content Management Systems) were rolled into one.
  • Cost was not so much a priority as efficiency.
  • 300 applications had to be replaced.
  • 50 people did the migrating work.
  • Only 10% of machines needed to keep that other OS as no simple replacement could be found for some applications on GNU/Linux.
  • To date 6900 PCs have been migrated.
  • By 2013 it is expected that 12000 of 15000 PCs will be GNU/Linux-only.
  • Heavy use will be made of web-applications to facilitate future changes to the system.

See 20 years Linux
How the Penguin came to Munich

On a lighter note, a councillor asked hard questions about going back to that other OS but the mayor promised to stay the course.

So the FUD about the Munich Migration being a disaster of some kind is not real. They are doing much more than migrating to another OS at the same time by rationalizing IT in other ways (CMS and web-applications) and they are not in any hurry crunching 50-100 per week and likely doing some hand-holding at the same time. It works for them.


  1. oiaohm

    oldman nice. Windows 2003 Enterprise and Windows XP 32 bit SP3 and later in ndis(network stack interface for drivers) are 100 percent identical.
    Note SP3 and later. SP2 and before they are not 100 percent identical so the Dell person might not have been a complete idiot just lacking the fine details.

    Its close enough that the driver works but 2003 drops out about every 6 to 8 hours long enough most people don’t notice the problem unless of course you are trying to really run a server with it. Remote areas laptops power usage ends up server so more hardware can run off solar. This is also why some people started suffering from strange network problems after SP3 was installed into XP.

    XP x64 and Windows 2003 x64 are 100 percent the same kernel in fact. So yep hardware is x64 and you have XP x64 drivers install 64 bit version of 2003 everything is good.

    I have had a few dells and be pain and require the Dell cut of Ubuntu so wireless works. Ubuntu from Ubuntu main site fail. Mind you these were the more expensive Precision Workstations that are tested with Redhat enterprise. Those work perfectly with Fedora and Debian and everything else Redhat related but would go stupid with Ubuntu from Ubuntu.

    Basically a Dell is a mixed beast. Interesting to know what series group yours was if you had trouble with fedora or debian on it.

    Yes the wording –”n-Series” systems ship with FreeDOS on a CD in the box.– Yes you order a n-Series with FreeDOS what you really get is a computer with a blank harddrive with a FreeDOS cd for you to install yourself.

    Yet for some reason you cannot choose to not be sent the FreeDOS cd.

  2. oldman

    “oldman I have had Ubuntu fail on wireless and Fedora and Debian work. Ubuntu was a bad mistake wrong firmware for kernel driver.”

    It worked fine on my Dell out of the box.


    YMMV. Mr. oiaohm.

    In the meanwhile I was able to run windows 2003 Enterprise edition on the same portable using windows xp drivers to fill in the support. Something I was told by Dell said should not be possible.

    Needless to say the person who saidf this was an idiot.

  3. oiaohm

    “Munich IT mayor Christine Strobl (SPD) said that the client, version 3.0 of which has since been released, has made it necessary to switch distributions from Debian Etch to Ubuntu 8.04 in order to better match the council departments’ requirements”

    Contrarian that is not translation error as such as minor context error people fall for like oldman. Debian Etch and Ubuntu are/were not used raw. At the time Ubuntu had more modern software than Etch. If you had read more of the documents you would have been aware that its not raw usage. Version 3.0 what version of Ubuntu is version 3.0 there is no such thing. LiMux is a distribution in it own right with own version numbers there are latter versions. The use a nice little tool to make it by the way.

    So they changed the base of there distribution. Mint has changed from debian to ubuntu back to debian. over the same time frame.

    They have altered ubuntu so they are not using stock. There images have pulseaudio and other things missing that are common trouble makers. Ubuntu does not work out the box for Munich.

    That ubuntu cannot be used stock for deployment and items due to stability have to be removed this is very much a condominium of Ubuntu. Reason Debian was used stock with things added not removed.

    Basically Debian stock was stable enough to work but software was too old. Ubuntu had more modern software but had to be fixed for stability.

    No large deployment case has used Ubuntu stock. This is a reason why I don’t recommend it to new users. If it is a quality desktop you should be able to use it stock. If I have to alter it to make it stable it is not worth a cracker to me. Ubuntu is simply over hyped.

    oldman I have had Ubuntu fail on wireless and Fedora and Debian work. Ubuntu was a bad mistake wrong firmware for kernel driver.

    Contrarian german government of course using openoffice in schools they are training there next lot of staff. Also not like it really matters if students use MS Office. Microsoft changes there interface about once every 5 years and it will take longer than that for kids to get through school so they will have to relearn anyhow. If OpenOffice/Libreoffice reduces relearning it will give students more learning time to learn more important things.

    Also OpenOffice/libreoffice as the dominance in schools is MS worst nightmare. Since when someone wants to write up a short document they will not ask for word but for free writer and so on. So reducing MS Office sales. I knew there was a current road map of the desktop side. With all the corrections for historic stuff ups on it.

    Clearly shows when the OpenOffice migration stops and the Linux Desktop Migration starts. Reason why have been ripping into you for your use of numbers Contrarian. How can the 5.9 billion apply to Linux migration when most of that time Linux migration was not the project.

    Also the last 3000 are the more complex cases there plan does include the possibility of using emulation. None of the 12000 will are using emulation of any form wine or virtual machines. This is why 100 percent still might be hit. There own documents state emulation may have to be used.

  4. Robert Pogson

    Nonsense. The hardware is handled by the Linux kernel and its drivers. Canonical has been criticized for how little they have contributed there. They certainly tweaked higher-level stuff but Debian GNU/Linux has given me little trouble with random devices wirelessly. Last year, we went wireless at the peripheries of our LAN and the COTS PCI cards and various notebooks encountered had no problem.

  5. oldman

    “GNOME is more friendly than GNOME?”

    Lets put it this way Pog, Ubuntu was the only distro that could at least in my experience manage pc hardware without any manual help. It was the only distro that handled wireless properly.

  6. oldman

    “I don’t find the Ubuntu treadmill much more comfortable.”

    But a lot of people do Pog. The reality is that for all its flaws Ubuntu has been the only commercially offered non geek friendly version of a personal linux desktop on the market. What I find interesting is that Munich chose to go with 8.04. Presumably it is because this version of Ubuntu is viewed as the most recent stable version that can be used.

  7. Robert Pogson

    Last time I checked, Ubuntu was based on Debian GNU/Linux. They could probably use Debian Testing for their purposes. I don’t know why anyone would choose Ubuntu over Debian when they were familiar. I chose Ubuntu once and regretted it. I got off the Wintel treadmill for a reason. I don’t find the Ubuntu treadmill much more comfortable.

  8. Contrarian

    BTW, it looks like even Munich has come to the conclusion that Debian is not for them. The article above states:

    “Munich IT mayor Christine Strobl (SPD) said that the client, version 3.0 of which has since been released, has made it necessary to switch distributions from Debian Etch to Ubuntu 8.04 in order to better match the council departments’ requirements”

    Better check with #oiaohm to see if you can claim that something has been lost in the translation.

  9. Contrarian

    “In 2009, Florian stated that all the schools used ”

    But they used it on Windows, #pogson! Are you forgetting the recent discussion sparked by:

    Too bad that the Munich taxpayers are too stingy to give the kids a leg up on learning MS Office, though. It shouldn’t take them much time, though, OO has copied almost all of the MSO stuff and moving to the real thing when they get a real job should not be too hard to do.

  10. Contrarian

    “The IDC information is really off topic to here. Also that issue was part of the reason you had never paid for a full IDC report and read the methods sections. Yet you dared to use the numbers”

    Are you saying that you did pay for the IDC full report, #oiaohm, and therefore know better? No one with such a trivial understanding of markets and sales as you have would ever do such a thing. You are a poor liar although you seem to work at it constantly. Of course the IDC information is not on the current topic outside of its proving that you are a fool. You made the statement that you understood finance and marketing, but your previous posts prove that is a lie.

  11. oiaohm

    “50 people 7 years to convert the first 6900 workstations?”
    If fact it not in a single report Contrarian. Because that 7 years is broken down into sections. Each covered by a reports. There is not a constant number of staff for 7 years working on Desktop. The reports over that time frame cause you problems.

    Last reports for 2010 on states the staff being used in migration what is 50 of course not full time on the project they have other maintenance work as well. Reason they are the general IT staff not some special department.

    There is a state of 50 for 2007-2009 reports. But in that time frame they were doing many things. Conversion from MS Office to Openoffice part time. Also the section of the staff 2007-2009 were working on new web interfaces to replace legacy techs. So not all decanted to Linux Desktop Migration. These are not really working on migration to Linux but clean up work. Still coming out of the LiMux budget.

    So unless people can be in 2 places at once 50 working at once on Linux desktop migration is not possible for any time in the project. I guess you are a person who believes they can be in two places at once or using figures without enough information to understand them? Contrarian.

    For the 2005-2006 trial there is no state the 50 IT staff being used. The trial did not use the complete IT staff. This is also stated in the reports for the trial.

    There is no staff work other than a 5 people before 2005. I don’t know where you get the took 7 years with 50 staff. took 7 years is bogus with 50 staff. There are only fragments of that time would the staff would be in use on the project in the sections related to Desktop working on migrations.

    You missed read what was said I would say german has nasty cultural traps compared you what you know. “50 people did the migrating work.” You see this a bit. Never mentioning full time doing it. Germans don’t mention full-time presume part-time. Cultural difference. USA don’t mention anything presume full-time so completely different context based on your background. Its a mirror. Yes 50 part-time people due to them doing other work as well would be the way to read it.

    5.9M euro project over run on what. Several new web interfaces that were not in the project document have been built and came out of the LiMux budget. Is the Linux desktop migration related or is this simply cleaning house processes that should have happened anyhow.

    We know a section of the 5.9M directly is linked to MS Office to OpenOffice. This is an expense someone could run into just migrating to OpenOffice on Windows. Is the really a Linux Migration expense. It is iffy. To be correct this is a OpenOffice migration expense. That if they decided to stay with windows they would keep. Yes really hard to call this a Linux Desktop Expense. Issue we don’t have the exact cost figure for this.
    Yes its the June 2010 report when restarted work on active migration that the number is stated. That you have not read Contrarian. Target for completion is 12000. Nothing says they have to stop 3000 short. Just the project is successful at 3000 short. In fact that report has a lot of useful information. A no point do they say 3000 short is because Linux cannot cut it just its the line in the sand they drew. 12000 machines have been chosen for 100 percent sure migration basically.

    Also the June 2010 states that the current 15000 machines are temporary and had been migrated to open source applications late 2009.

    German logic Google translate did a reasonable job getting it to english.
    “Last but not least is – as in every long-running project – that it is successful if it delivers what is currently needed and not what the parties have intended to project start.”
    I can read german but rewrite it as english I am stuffed you have my dyslexia effect on my english. combined this with translation you really don’t want to see it. So forgive the use of translate engine otherwise I would have had to quote the raw German.

    This clause has been on almost every LiMux report. The class of successful is zero disruption to operations.

    It is important because it explains why the project will overrun and the overrun will be accepted. If something happens that need attention so current operations work if it was in the plan or not LiMux has to fix it. Yes why it a serous question 5.9M euro overrun on what. Linux Migration or Website repairs or some other bright spark idea to provide a new interface for something… Basically the LiMux budget is a catch all budget due to its mandate.

    Basically stop acting as if LiMux spend money is just for Linux Desktop Migration. Yes the project stated goals that you have never read Contrarian or failed to read. Wikipedia perfect english no excuse here since you need no german skill to read it “The decision in 2003 had two components, on the one hand to get free software running on most of the desktops, and on the other hand to buy and develop web-based and platform independent (e.g. Java-based) business applications.”

    Note three things its for Most desktops not all and that its two independent projects in the one budget. Web platform and Desktop. So what one overrun for 5.9M by how much thinking they are both billed under LiMux. Do you have the figures. I know I don’t Contrarian its nice to kick the Linux desktop with anything you can get your hands on even if its not solid numbers.
    “get free software running on most of the desktops” Yes this is a very lose victory define. That they have Openoffice, firefox and thunderbird on all the desktops by end of 2009 they were successful to meet the project objectives.

    The 80 percent goal is also stated here it also explains why the Foreign Office went south so baddy. Again in german.

    Yes also contained handy information like they were using Microsoft Office 2000. Now who wants to retrain Microsoft Office 2000 users on Microsoft Office 2010? From a user training point of view libreoffice/openoffice is closer.

    The IDC information is really off topic to here. Also that issue was part of the reason you had never paid for a full IDC report and read the methods sections. Yet you dared to use the numbers.

    Your incompetence with numbers and research keeps on coming up Contrarian. You use numbers with no solid base about time you stop that. 5.9 M euro is not a solid number to be using. Since you don’t know what the over run was on.

    I have been able to state where some overruns were from the exposed reports. But that is not all of them.

    Basically if you cannot read german due to all the key LiMux source documents being in german you really should keep your month shut. Contrarian otherwise you are going to keep on goofing up badly and annoy us who can read german. Reading an english translation without understand the culture as well will cause you to miss understand the documents as well.

    Simply a lot of people commenting on LiMux really don’t have the skills to. Contrarian you are not alone in not knowing this.

    Heck you could not read the wikipedia english page to get a basic background on what the project covered so I should I expect this level of incompetence from you Contrarian. Or is this going to be a one time mistake.

    Basically LiMux is too murky to be used to attack Linux Desktop migration. Can be used to prove viability of Linux desktop migration as a TCO base but that is about it. The murkyness is ideal for a worst cast TCO.

    Yes the murkyness of LiMux is why I am tell you that you have to TCO the complete project. You cannot get enough information to make 100 percent clear cash reading on over budgets and other things.

    All you know they spent more. Not on what. Was it refunding the project for other work that was required?

  12. Contrarian

    “That is not a failure of GNU/Linux in any way. GNU/Linux could run the application if it existed”

    I would argue that it is a failure in the sense that it makes Linux useless for the individual involved.

    Poring through the above link on this thread, I noticed that it claimed that there were some 1000 IT jobs associated with maintaining the 15000 workstations involved overall in the project along with some 28000 computers in the school system. That seems excessive in any scenario, eh? It may be that there is so much featherbedding going on in Munich IT circles that any initiative at all would come up looking rather inefficient. Also, it was noted previously elsewhere that the educational machines are not involved with conversions to Linux and the schools have opted out of the project.

  13. Robert Pogson

    Contrarian wrote, “Where does it not say that, after all is said and done, 3000 of the 15000 workstations will remain as Windows since Linux cannot cut it?”

    “Facts and Figures

    as of March 2011

  14. 15,000 jobs to use free software such as Thunderbird and Firefox
  15. 15,000 jobs using and the wiki
  16. Nearly 6000 people use the LiMux client (either Debian Etch or Ubuntu 8.04, KDE 3.5)
  17. 10 of 22 areas of migration are already migrated
  18. Target by the end of the project

    Migration of at least 80% of the administrative client PCs on the LiMux”

    So, the site reports 80% of 15000 PCs is the least conversion rate. That is, it could be more. That is quite different than 3000 will remain unconverted. It could be 90%, leaving 1500 unconverted and it says nothing of why the conversions might be remaining undone. In this article it is stated, “Even politically, the project set up seems to be still solid. Long-term goal, is converted by 2014 80 percent of all computers on LiMux to have and it may now shake probably no one. “There are simply cases, there is a switch to Linux, not just economically,” said Maier.”

    So, the problem is that in a few cases, migration is not economical. e.g. One guy runs one application. It would be wasteful to write a new application for so few users. That is not a failure of GNU/Linux in any way. GNU/Linux could run the application if it existed. Of course, the ISV may eventually see it in his interest to port to GNU/Linux.

  19. Contrarian

    “Sorry I know finance and marketing. Law to my regional area.”

    My feeling is that you do not know crap from Shinola, #oiaohm. BTW, I am still waiting for you to come up with any sort of explaination for you bizarre comment about server market share, i.e.:

    ““The report contains two sets of revenue numbers. Factory and Customer.”

    #oiaohm, you are such a transparent BS generator, it defies belief. I am intrigued, though. Could you please provide a definition for “Factory Revenue” and contrast it with “Customer Revenue”? I cannt imagine what you might try to say about that and I think it would be fascinating to see you try.

    Put up or shut up, fool.

    “Do you read german? Contarian I guess not. Because it would explain why you are being so much of a idiot here.”

    Your problem is that you seem to not understand English, #oiaohm. Where in German does it not say that it took 50 people 7 years to convert the first 6900 workstations? Where does it say that the cost overruns are not 5.9M euros? Where does it not say that, after all is said and done, 3000 of the 15000 workstations will remain as Windows since Linux cannot cut it?

  20. oiaohm

    Contarian simple fact of the issue if you read german you will find that stated clearly in the LiMux reports is that the costs are inside yearly network maintenance costs of the old system.

    Sorry I know finance and marketing. Law to my regional area.

    Do you read german? Contarian I guess not. Because it would explain why you are being so much of a idiot here. So are unable to understand how the Munich TCO breaks down.

    Cost of the Linux migration is less than paying the Microsoft licenses also about the same as maintaining a Microsoft based solution. Yes lower maintenance costs have been reported on Linux as well when not in migration. You don’t have those strange machines from time to time that when a windows update applies they go south and other issues that are windows only.

    Simple fact Contrarian you are defeated you don’t want to accept that so you have to insult me. You have not done your homework on .net solutions I have.

    Most people don’t understand the downsides of database connection pooling. Removes the means for each connection to be a independent user. So making the risks of sql injection worse. Why must you connection pool in Windows the network stack limits to prevent cal abuse will come back and bit you otherwise.

    Simple point here Contrarian I like doing secure systems. Have my hands tied by a CAL licensing system protection that is not working for me. That I have to watch out and avoid all the time.

    Most likely Contrarian you don’t know enough why you cannot do things under Windows Servers that you can do under Linux and Unix servers without issues and that a lot of it traces back to the CAL licensing system protection methods. CAL licensing system protection methods are basically a boat anker. Yes why can Linux run huge number of databases connections without connection pooling and windows kind requires it or sites fail? Was one of my early questions that got me hating CAL enforcement systems.

    Yes I have messed around with reactos and yes some things you can do under reactos( even that it still highly crud) will not work under windows so yes the issue is the CAL licensing system protection methods not the windows NT design itself.

    Even Windows 7 home has some protections in software over license over 10 machines without a server and machines start magically appearing and disappearing. Patch the right file and everything will just work.

    My disgust at Microsoft Windows comes from doing homework and finding out exacty why things I tried todo with it did not work. CAL system inside windwos is a prick. So just because one bit will not charge you CALs. The CAL enforcement system is sitting in there like a land mine waiting for you to make one mistake.

    I have seen companies trying to desktop multiply windows XP as well to 20 + seats. Yes they run into locks. Same locks you found in 2003 and 2008 for CAL enforcement.

    I guess Contrarian you think .net website are ok because you don’t really know Windows. So are not properly disgusted with the idea of running a webserver on it due to the extra set of issues you are exposing yourself to. That a simple minor alteration might take site off line because something in the CAL enforcement system took offense.

    I don’t like the idea of playing with land mines. I either want them fully disarmed or no where near me.

  21. Contrarian

    #oiaohm, you seem to know very little about law, marketing, finance, or much else in spite of your labored discussions on miniscule IT admin operations. I suspect that you got that by copying something off a web page, but I don’t really care. Your clumsy wording nothwithstanding, your anecdotal “evidence” is just pure hot air.

  22. oiaohm

    You also missing the critical points Contrarian.

    With .net website You have never done it for large numbers of people Contrarian.
    Did you not think that using Linux network connected storage does not come back to bite you as well. The issue you run into Contrarian is the way MS has done there disk access/network and cal enforcement. Yes I know the disaster you are describing as solution. Ie we will use .net website with connection pooling. I had the luck of that being a place with 1000+ users at once. Lets just say the MS cal limits clearly showed themselves. First hand tells me go Linux with web server way less pain and risk. Forget .net. Java and php plus maybe spend some money on oracle database.

    Do you think running out of date software is healthy? Contrarian

    When you look at the global numbers MS income is too low for everyone running upto date software by a large margin or there has to be a hell load of piracy web number back up lot of people running out of date software.

    The Munich success story is way more affordable to get people on to current versions of software than the Microsoft path.

    Cleaning up virus infections and the like. All costs.

    Also you are stupid not to listen to what I said.

    If you had done up a full TCO of the Windows solution you would find that most of the cost of doing Linux you are paying anyhow in support staff.

    Of course Contrarian we were using first world figures the complete 1.5 billion machines are not in the first world. You did not take account of there current maintenance costs.

    When you don’t bogus the numbers its quite doable. Since that amount is being spent anyhow keeping the systems running.

    The conversion to Linux turns up in a downturn for the simple fact how much it costs to maintain a computer system is the same money to convert in lots of cases. Contrarian.

    Linux is not a straight drop in but the change over can be done without major disruption if it planned well as part of general maintenance.

    The simple point here Contrarian the idea that is not practical proves to me how little you know about what the real world is spending keeping all this old windows software running.

    To be correct TCO compare you need todo 3 numbers.

    TCO to install Windows TCO to install Linux TCO of maintaining the OLD SYSTEM.

    Munich it was out the bugget to afford going up to new Windows. Cost of maintaining old system don’t exactly get better with age. Microsoft particular does see to this with update causing older machines to lose performance to remain secure. Linux TCO works out about the same as the old system.

    Basically if you were not being foolish and did what you told todo for once Contrarian you would have worked out your mistake and not have to have me directly tell you.

    Munich maintenance costs are way lower on all the Linux machines to stay current for ever. Costs are Costs. Contrarian. You have the magic idea that € 5.9 million would not been spent anyhow.

    Since the 5.9 is labour different areas of the world the price alters majorly.

  23. Contrarian

    “To be correct it did not”

    Why am I not surprised that people you associate with are stupid, #oiaohm! There seems to be a unifying factor at work. :-)

    All the design tenets need to be followed, of course, putting business logic middleware near the data sources to minimize the data that has to flow to the client. If you leave some anachronisms around, you might very well need to pay for a few more CALs to be legal, but if you have a savvy supplier, you won’t have to do that and your system will work much better for it.

    “Also if you do the maths on the 1.5 billion machines globally …”

    Again, you miss the point, #oiahom! Par for your course, of course! :-)

    #pogson calls the expenditure of so much money with so little return as a “Munich success story”. I am merely pointing out that such an “economy” would bankrupt everyone if they applied this “success” to the world. You come along and naively point out that it is not practical, giving me a chuckle, and making it necessary to explain it all to you.

  24. oiaohm

    Contrarian I have a nightmare for you. I had a group using web-services idea avoid CALs. To be correct it did not. Reason they were accessing network shared disk. Windows 2008 server allows so many connections/users to network storage based on Cal’s paid.

    Cure was either install Linux for network storage or pay MS Cals 2008. Also I have seen people caught with terminal services this way on windows as well. Ie paid for terminal services CALs forgot to pay for 2008 server cals because everything would be running on the server so it broke.

    “CALs are not needed for web service connections, period.” Yes this is bull crap Contrarian. You don’t need cals for the web connections but you do need cals for what the .net application running on the web server is connecting to. So a web service on windows can turn down right expensive fast.

    Yes there is a reason why businesses end up running Linux as their web servers. No more Cal nightmares breaking stuff.

    Contrarian I really do fix up these deployments I work with MS Cals daily. I hate them with a passion due to the number of ways companies get hurt because of them. Its ok if I buy the Cals. I know how they are interlinked. Problem is when it gets up to management of many companies they start thinking we don’t need that so don’t pay for key Cals they in fact depend on because they are so expensive. Then there networks develop issues because of lack of Cals.

    Contrarian also Munich example is basically the worst case. Also Munich is rare because most other examples like the french policy the migration to OpenOffice and the Migration to Linux were done as two independent projects.

    $10M in real money is less than paying the Microsoft bill Contrarian on MS Office and its hangers on. So by 2010 with the 100 percent conversion to OpenOffice they are already in profit.

    Did they miss the mark of cost saving no. That was 5 years from the start of work.

    “If it took 50 IT personnell 7 years to migrate 6900 computers” Be at least half aware here.

    15000 computers after 5 years work running OpenOffice thunderbird and Firefox. Not 7. First 2 years were wasted in political infighting coming up with any reason why the open source option could not go forwards.

    The migration to OpenOffice if you want to get to real numbers only took 3 years after the trial. 2 years longer than expected.

    Basically cost saving have been achieved Contrarian converting to Linux is icing on the cake.

    The answer is 5 million staff 3 years to get 50 percent + of the savings for the 1.5 billion machines world wide. Yes just getting rid of MS Office saves you money. Linux conversion after that is about 4 years same staff to pick up the extra savings.

    Also if you do the maths on the 1.5 billion machines globally a lot of people cannot be updating MS Office and its hangers on for business usage or MS should have income in the trillion totaled up over 5 years. About 5 trillion USD over the 10 years if everyone was updating as per Ms life cycle with the MS complete set just in licenses. MS number reporting on MS office is not even close to that. MS does not have a 1/2 trillion dollar a year income. So we know there are lot of people not on the MS cycle.

    1.7 trillion is still cheap Contrarian to convert the Lot even that its basically double the real figure yes you can bogus the number and you still lose. Simple fact is you did not do the maths of updating the lot equal level with MS product. So yes your figure turns into about 4 years paying Microsoft todo it when its really about 2. This is a constantish figure. Its about 2 years paying to stay up with MS products to get off MS products at worst.

    Bigger you make the numbers the worse MS looks Contrarian also how poor MS is performing comes very clear.

    Next time you want to say a figure on this do yourself a favor do up both figures. The MS figure and the Linux figure using the same metrics.

  25. Robert Pogson

    At the time of the migration, Munich was using a ton of servers and services many of which needed CALs with XP/2003. The cost of changing all their applications was a part of the cost of migrating to GNU/Linux. Would you have had them invest all that effort and ended up with XP needing upgrading to “7″?

  26. Contrarian

    What does XP have to do with it, #pogson? CALs are not needed for web service connections, period. .NET apps consuming web service connections using on XP are no different than apps on Windows Vista or Windows 7 or Windows 8 in the future. You are behind the times.

  27. Contrarian

    “, alone, can convert 100 machines in a few hours…”

    Perhaps you can, but then there is Munich. If they took their time and collaboratied with one another and came up with the scheme that they have been implementing for the past 7 years, what can you say? They have spent some 8 million euros, over $10M in real money, and come up far short of their original mark. Are they all fools? Or are you grossly understating the effort needed?

    “As if that other OS were not an ancient thing in IT.”

    But not as ancient as the terminal and server architecture. Furthermore, no CALs needed. You suggest there are savings to be made with Linux when the costs that you are using are easily avoided with more modern design. Your argument applies to a long gone age.

  28. Robert Pogson

    Contrarian wrote, “If it took 50 IT personnell 7 years to migrate 6900 computers…”.

    I, alone, can convert 100 machines in a few hours. I’ll work for food… Cost is about $1 per machine. A bargain, I would say, in light of the many benefits of using FLOSS.

    1. Plug the GNU/Linux terminal server into the LAN.
    2. Kill the old DHCP server.
    3. Plug in thin clients booting PXE or set BIOS of machine to boot PXE.
    4. Repeat step 3 as often as needed.
    5. Enjoy.
  29. Robert Pogson

    Contrarian, writing of my love of GNU/Linux thin clients wrote, “You are stuck on an ancient architecture”.

    As if that other OS were not an ancient thing in IT.

    Thin clients can work a number of ways. Some come with a browser and you get a web interface to some portal straight away. Some use X or RDP to connect to one or more terminal servers. It’s all the same getting the most value from a few machines and spreading the joy over many. It’s an efficient means of operation and still valid today. It is perfect for school labs, for instance. The teacher can “kill” and or all students with a few clicks. That gets attention/respect/whatever and allows the teacher to make the best use of time in using IT. Instead of having a highly-paid educator watching a clock, the server can shut the lab down at the right time preventing straggling and so on.

    Electricity is old. Copper is old. Flowers are old. They are all valuable.

  30. Contrarian

    “They would have needed CALs too so we saved $35000 just on server licences.”

    I think that you are working in the stone age of IT, #pogson! You fuss about with “thin clients” and perhaps there is an issue of CALs involved there, but if you were to simply think “web services” and WCF, all that goes away and you are in the modern era. How do you think that phone apps work? They are not wired to servers and CALs. Neither are tablets or netbooks or regular PCs these days. You are stuck on an ancient architecture and apparently cannot see the woods because so many trees are blocking your view.

    One other thing that I noticed, since you are so fond of extrapolating results given a single data point:

    If it took 50 IT personnell 7 years to migrate 6900 computers using the “successful” Munich technique, how many would be needed to convert the whole world? The simple math, assumming your 1.5B count of PCs in service, shows about 11 million IT weenies working until 2018 would be required. I won’t bother with trying to adjust that just for the USofA, but it would dwarf the latest scheme from the president to make work for enough people to get the economy rolling again!

    If we use the Munich budgetary figures, apparently some 8 million euros for the 6900 computers now converted, we get about 1.7 trillion needed to get the Windows monkey off our backs.

    What other success stories do you have to tell?

  31. Robert Pogson

    ch wrote, “It was a purely political decission. Why should any company – or any public administration trying to save money – follow that route, given the cost ?”. I have migrated a bunch of organizations and the lack of purchase-order requisitions and draws on petty cash belie “costs”. The chief cost was my time which was already on the payroll and in fact my time spent on IT was reduced by migrating because GNU/Linux is so easily managed and has so few malwares.

    Here are several reasons given by the French national police:
    “The reasons behind the move are tri-fold, explained Geraud: first, to reduce the force’s reliance on one company and offer more choice by diversifying IT suppliers; second, to give the gendarmerie control and oversight of the operating system; and third — cost. This last might have been reason enough all by itself. The move away from Microsoft licensed products is saving the gendarmerie about seven million euros (10.3 million dollars) a year for all its PCs.”

    They sound good to me and they are not about politics but performance.

  32. oiaohm

    “That was of corse not the price for the licenses, as you alluded before, but the price for the whole project.”

    Ch fact please don’t lie. MS quote only covers MS part of the Project not the whole project. Ch about time you dig out the documents what the MS quote covered. The MS price did not include on-site deployment costs since that was meant to be done by Munich existing staff. Since MS would not be providing that. Also shock horror the MS price did not include anti-virus. Also the first has a repay at 5 years for the next version of MS products. Ie $36.6 now then every 5 years another $36.6 to remain current. LiMux price tag is not even close to what MS quoted.

    So how is it the whole project when the staff doing the work are not in the quote neither is all software required.

    All MS covered was the MS licenses and the on-line support for the second quote and a repay for next versions at 6 years.

    So LiMux price and MS price is not apples to apples. MS price is under quoted by a huge margin.

    Ch you have not answered how you thought 400 dollars was possible. If you wish I can do a detailed list how to 14000 machines can end up costing $36.6 million USD in licenses. Its only 2615 a seat. Simple to achieve.
    Clients Windows+MS Office enterprise
    Server Windows+exchange+sharepoint+lync(had a different name back then Microsoft Office Communicator)+Project Portfolio server+Performance Point server.
    That gets you to 2615 per seat or more. Of course you have duplication on lot of those cals for backup servers.

    Issue here ch large deployments are something I have sourced for. I am use to the fact at times a 500 dollar machine of hardware may have 2500+ of Microsoft software on it. Lots of that hidden in server side cals. Part of the reason why I see cals as evil. They hid the cost per seat quite well from lots of people.

    “It was a purely political decission. Why should any company – or any public administration trying to save money – follow that route, given the cost ?”

    What does the fact a political decision have to do with it. If I am looking at it for cost savings. I am purely look at how much Munich spent and how much I would have to spend or have spend doing MS solutions the same. In fact for cost saving assessment that sections went pair shaped makes it even better. Since it was not a ideal run.

    First one todo never knew the cost saving would exist only suspected so is a risk. Following know the cost saving exist. Also anyone like me knows what the real costs are know that what Munich has done so far has cost them less than spend on a MS solution. We are watching to see by how much. Yes given the cost is why considering to follow that route is possible.

    Yes it going to get simpler in 2013 when Munich can documented the complete projects on going cost savings and the fact of lack of over spend compared to MS solution. Also Munich fixed up a lot of issues no one else has to battle past. The first always has the hardest path.

    Yes given the cost now a proven path as long as 80 to 90 percent is done in 2013.

    Yes given the cost is why they should consider it. Munich is getting out very cheaply.

    Ch your german is either rusty or your minorally miss read context of the wording due missing context information of other documents that stated the number of instances of CMS as 22. It cannot be 7 as you read more documents. 22 departments were running 22 CMS instances made up 7 different CMS systems that were incompatible with each other. Yes its was 22 instances to 1 instance and 7 differently designed systems to 1 system. Different documents mention the number independent instances.

    “7 different(configuration/makes of) CMS installation(s)” is the only way that section works with the context of the other documents. Yes the section in the source documents is bit ambiguous.

    ch yes I read German. This is how come I said before that the wikipedia numbers for machines are correct but the english project page for LiMux are not. That you can read German its about time you start reading all the source documents so you don’t make any more context errors or exaggerations. Yes say the MS quote was the whole project is an exaggeration. Since the quote did not contain many things.

  33. ch

    “Reason for buying expired versions of Windows it was cheaper.”
    Even worse. Anyone in an IT department suggesting that needs to be fired.

    “Reality a Windows network almost never ends up 1 desktop OS end to end.”
    Can we at least agree that the differences between WinXP, WinVista and Win7 are _much_ less than the difference between Windows and Linux ?

    Regarding prices, may I just point out that Office (without Access) _retails_ for €180, and for 15,000 seats you will get some rebate ? And I tend to doubt your €50 / year for anti-virus. At least the articles RP has quoted make it clear that the decission was _not_ primarily about saving costs.

    BTW, any company contemplating switching to Linux is very likely to tend towards RHEL or SLE* (at least on servers), and they are not for free, either.

  34. ch

    “In the first instance, M$ reduced their price to compete and it was multiples of €5.9million.”
    That was of corse not the price for the licenses, as you alluded before, but the price for the whole project. So:

    “Munich was not just about price. M$’s reduced price was less than the plan that was approved in 2004.”
    It was a purely political decission. Why should any company – or any public administration trying to save money – follow that route, given the cost ?

    BTW: In other comments I noticed a bit of confusion about certain numbers. I happen to be a German, so I can tell you that your sources actually say:
    The city administration of Munich has 22 main departments that before the project used 7 different CMS installations. As one result of the project, those 7 CMS were rolled into one. There were no departmental reorganisations as part of the project.

  35. oiaohm

    Good link Robert Pogson. $23.7 million is still more than running LiMux for the 10 years. That is with a 35% discount. Remember the 23.7 million from Microsoft does not include paying the on-site support staff. Yes once you add on the on-site support staff for the machines you are behind by a decent percentage.

    LiMux budget is everything. Software and on-site support staff. Microsoft was throwing in free MS phone support. Not that useful really.

    Simple fact ch and everyone else the maths are on Munich side. Everyone claiming over budget about it don’t a have a single clue what they are talking about. Its not over budget when you compare to the costs of the MS option.

    As long as LiMux has the 12000 or more sometime in 2013 its 100 percent viable for them. They will skip the next upgrade cycle the Windows 8 upgrade cycle completely for over 12000 machines. The remaining 10 to 20 percent will be viable to put 10 to 30 staff to sorting out what ever issues are in the way. 100 percent Linux might still happen there. Last 10 percent is always the hardest. So the last 10 might take another 10 years. But does that really matter when the cost savings in the 90 percent is more than paying staff to sort out the 10 percent.

    Even if government in Munich changed now. The replacement would have to look at the books and basically go darn cannot afford to change path. MS is going to be more expensive so it has to go. Remember the most expensive part of the migration is already done. Breaking MS Office hold.

  36. oiaohm

    “- Because it took so long, Munich had to buy new Windows and Office licenses anyway. (They started with PCs running Windows NT.)”

    Incorrect information I don’t know where you got it. They did not buy anymore Office licenses. OpenOffice migration was underway at that stage. Result was the MS Office license displaced by OpenOffice was redeployed in combination with the extra windows licenses.

    At the start not all there machines were running NT. Some were on windows 2000 and XP before the start of project test pilot in 2005.

    Now the extra copies of windows aquired were not new. Good look at o my god. Yes expired support versions of Windows is what was aquired so staff did not need training. They never bought new licenses of Windows second hand yes. Zero extra copies of MS Office.

    Reason for buying expired versions of Windows it was cheaper. If I was the IT in that place I would wanting migration to Linux as fast as able just to get rid of the secuirty risk of running all that old software.

    Also read the date they aquired extra licenses. Remember the test pilot over run before that was caused by FUD interference. It was never expected for the migration to Linux to be fully under way by 2007. But it was expected to be a little more advanced if pilot had not been messed up it would have been.

    “- Only 90% of PCs can be migrated, so you wind up supporting two desktop OSes instead of one.”

    Reality a Windows network almost never ends up 1 desktop OS end to end. Always ends up fragmented new machines mixed with old normally normally cause this. When you look at there windows network NT mixed with 2000 and XP was already very messy before the migration started.

    Result of the process looks like it will be 90% dependability on 1 Linux OS forever more. Reason free upgrades to current on all the Linux desktops. 10 percent mixed windows versions. This is far cleaner.

    Remember 2000 and NT did not have COA stickers but COA certificates. So they can keep on use those for ever more in virtual machines.

    “Just the additional €5.9m are ~ €400 per PC, more than MS licenses (Windows/Office/server CAL) would cost.”

    ch >>Really I would like you if some how your maths work for under €400 a seat. There is no way I can make it fly and run Windows Servers. Running Linux servers its still down right tight in fact I still blow the 5.9 million. Windows + anti-virus just. No MS Office.<<

    Here is my maths. I guess you forgot the hanger on and that the time frame was 3 years.

    Average price €50 a year for the anti-virus. 5.9 is over 3 years so subtract 150 from the 400 to start off with. €250 is all you have to play with for windows software. Now you should be able to see you are screwed. I am now going to spell out all options.

    Next exchange cal per device is €280. So anti-virus + exchange cals and you have nothing to buy Windows or MS Office with out of €400 for 3 years. So now lets presume they are using something else other than the exchange outlook combination ie thunderbird something just so we have cash for the client(error1). So what is the windows server cost just to connect a client €99 a seat approx. So that €250 has no vaporized to €150 for Windows and Office. Now you are screwed because you cannot even buy MS Office standard upgrade for under €150. Ok lets say OpenOffice works. Nop €150 is not enough to buy a copy of MS Windows new either that is meant to be hooked to a Windows Server. Now lets say we drop the Windows Server and run the server room fully Linux (error2). That restores the €250 and that is blown buying a professional copy of windows so its network friendly so still no copy of MS Office. Note average price for a professional copy of windows is about €350. So yes €500 a seat in licenses and I still don't have a copy of office. No office no servers and I have blown the 5.9 million budget for the 3 years.

    With office and Windows server and exchange cals for the 3 years you start going nice in the direction of €1000(and this is being nice the real figure is higher) a seat or 12 million or 6 years of budget LiMux project what really ends up about 13.5 million and a bit due to the extra anti-virus for the 3 years. Next at 5 years old with MS products you are meant to pay for Long term support. add in long term support payments. That takes you up 10 years Guess what LiMux running for 10 years costs less still. At 10 years for most MS products you are now end of life. So hello pay up again. Now this is when LiMux starts making a profit big time. Remember how much of the project has been one time costs.

    Remember the error1 and error2. Both error1 and error2 still need maintenance so have a cost. There is an error3 I have paid nothing for deployment or training or support yet.

    I am not kidding when I say 1 MS live cycle covers 5.9 million extension for a extra 3 years. Taking the total time of the project to 8 years or 10 years if you want to include the time when nothing major was being spent. It is in fact more than complete spend of the LiMux is less than the MS Products and Anti-virus hanger on cost for the same time frame.

    Way less remember MS is costing more before I pay support staff to support users operating it.

    Remember ch Munich government gets no MS discounts due to the fact they run some commercial parts to their operations. You are most likely thinking with education and government discounts that don't apply in this case.

    ch Second hand I might swing it but that is not secuirty equal. Really Linux vs running non supported versions of Windows. I think I would be going Linux.

    Before you say that is unfair over the anti-virus. Remember Linux can run real-time anti-virus scanning using clamav for free. Windows you don't have that option. So comparing equal to equal Windows has to pay the anti-virus and firewall bill.

    "a) Streamlining IT by weeding out among the gazillions of templates/forms and special applications etc. – something that should have been done anyway." This is in fact what took most of the time. Cleaner system Munich will have after this will look after them well. It was a huge one time cost.

    Now the fact that Munich is being called a disaster yet even with all the overruns it is still going to end up cheaper than staying with MS in license costs for the 10 years without the renew. Yes you even have enough in the budget to pay for the 2000 extra Windows 2000 licenses and still be ahead.

    Its really insane to be calling it a failure. Simple fact no one is doing the maths of how much Microsoft really costs. Its not cheap. Yes you can employ a team of 50 people working for you full time for 10 years for less when you are talking about 12000 machines. Just to be creative every 240 machines you can afford to employee one staff member for the amount you pay in licensing if you are paying what you should be.

    Now if your network is under 240 machines. Not considering Linux is a valid path. Over you should be since it means an extra staff member to take care of any Linux issues that turns up. You most likely have support personal anyhow.

  37. Robert Pogson

    In the first instance, M$ reduced their price to compete and it was multiples of €5.9million.

    Munich was not just about price. M$’s reduced price was less than the plan that was approved in 2004. The migration was about taking charge of IT and reducing costs and improving performance forever not just the next step on Wintel’s treadmill.

  38. ch

    First of all, the whole project eventually did two major things:
    a) Streamlining IT by weeding out among the gazillions of templates/forms and special applications etc. – something that should have been done anyway.
    b) Migrating to Linux.

    Look at the second aspect with the eyes of a manager, and what do you see ?
    - It takes ages.
    - Because it took so long, Munich had to buy new Windows and Office licenses anyway. (They started with PCs running Windows NT.)
    - Only 90% of PCs can be migrated, so you wind up supporting two desktop OSes instead of one.
    - Just the additional €5.9m are ~ €400 per PC, more than MS licenses (Windows/Office/server CAL) would cost.

    Not a very enticing notion.

  39. Robert Pogson

    Works for me. I have been using and LibreOffice for the last decade and had no problems getting work until this year. Not their fault. I am too old…

  40. oiaohm

    Ivan interesting phrase. Shows up as fine in “After the deadline”, “language tool” and Same with most grammar checkers.

    Reason its not a grammar fault. the base of link grammar that is in abiword is in fact checking for items out side the official define of grammar checking on english.

    RelEx is checking extra set of items. This includes triggering in statistically less common word patterns. This fall under linguistics not grammar.

    Its a nice double sided sword. Detect what you class as a fault Ivan could cause false positives with the variation around the world.

    Of course first generation languagetool would trigger on that grammar pattern due to its lack of tolerance to don’t. don’t and do not are regional.

    Also the stating negative outcome twice “don’t work” and “none” again appears in regional variations of english.

    Yes that Abiword detect it is a bug in some regional areas. Ivan basically little word of warning due to my Dyslexia I might type with what appears poor grammar. I have really studied english really heavy because the idea was that it would fix the problems I have. In fact it made them worse. Result was how much I found was just regional variations.

    Something I have to watch is that I write in the right regional variation of english. Listern to yoda in starwars. Everything he says technically fully valid english. Just its a mix match of regional variations of word orders. Yes I get told talk like yoda at times because I have mixed regional variations.

    Ivan the issue of doc formatting going to hell happens MS Word to Word even the same version. If you send out a resume you have not tested as Doc in word viewer on a different machine with different printers and it gets thrown it is your fault. Yes if the MS Word viewer cannot view it right normally it has something special in formatting or printer settings(yes I really do mean printer settings). One way a word to word falls apart is when you have locked something to page and word gets the smart idea of making that location based on printer dpi ie so many dots from top of page so many dots in. So when document is opened on a printer with a different printer dpi your document is shot to hell. Yes word does this random-ally and word viewer on you machine will get this right even that you document goes splat at the other end.

    How do you trigger this issue less random-ally leave only one printer in printers or all the printers in there have the same dpi. Yes the xps printer MS Office adds to the system is a hack to prevent the defective action that only part works.

    Writer on the other hand. Either an action exports right or fails. So you can make a nice list don’t use any of these features and the file will export perfectly fine.

    Basically Writer and Abiword you have they chance that the reason why your resume if it fails was your fault for not checking it. Compared to MS where it was some random evil that was introduced in MS Office 95 and has never disappeared. Reason for its introduction was to reduce the response time between pressing print and printing.

    Yes that dpi fault every time you save is rolling the dice with MS word.

    These days I would more likely send my resume as Pdf. At least that way I will not get eaten alive. Yes MS Office 2010 introduced PDF support for a reason.

  41. Ivan

    Also let us know when the phrase “This here grammar check don’t work none.” no longer passes the grammar check.

    Abiword flags this, making Writer substandard to other FOSS programs.

  42. Ivan

    Please let us know when OpenOffice and its poorly named fork have doc formatting that won’t get a resume thrown out.

  43. Robert Pogson

    oiaohm wrote, “the start of the Linux Migration got postponed many times due to the Document issues”.

    One postponement was due to political interference as well, FUD about software patents etc.

  44. oiaohm

    Phenom the point you keep on missing is the Linux migration stage is going as Clock work.

    22 CMS (Content Management Systems) were rolled into one. Was not 22 departments merged into 1.

    The 22 departments still exist. But they are all now contained inside the one CMS system so they now can share information with each other cleanly. This is an cost saving without firing people since it makes stuff more productive.

    Really the Project was quite well planned. Error in estimation for how much work would be require to convert the stock pile of documents. Office suite migration is the thing that is going to kill you.

    Being first means some errors. The deployment of Firefox and Thunderbird went completely to clockwork.

    Phenom the end date has only been moved once. From 2010 to 2013. This is not keeping on being postponed. Yes the start of the Linux Migration got postponed many times due to the Document issues.

    This is the problem about reading stuff out of context Phenom. Munich would have blown budget just migrating to Openoffice.

    I have seen Windows migrations blow out for exactly the same reasons. From Office 2000 to Office 2010 migration did not go pretty at one place to say the least. Again it was old documents with old macros failing to migrate stuffing everything up.

    Its the solid thing to take away from Munich don’t at any time underestimate how bad the archives of documents are going to bite you.

    Phenom really over Libreoffice.
    The releaseplan time table is completely done. Yes a officially documented road map is planned. .
    Phenom basically Libreoffice is young. Issue with printing up a roadmap is information. Yes they are collecting the information to write up a proper roadmap of feature with priorities.

    Also both Oracle and Sun neither did up feature roadmap and plans for OpenOffice. So basically the documentfoundation is working from scratch.

    They are currently running a vote for enhancement. This will effect the final roadmap.

    Also you over value solid roadmaps. Phenom. Having a solid roadmap is not even true for projects like KDE. KDE has a wishlist for each release. If a feature is not done its pushed to the next release.

    Its better done right than pushed out because its on a roadmap. Libreoffice is maintaining a feature lists and other items and attempt to achieve them.

    In time libreoffice will take on a KDE style model. But its getting there.

  45. Phenom

    Pogson, merging 22 departments into 1 will save costs only if people get sacked. That would save money, but it has nothing to do with the nature of software – proprietary or free.

    The fact that the project end keeps being postponed, and costs grow, tells two things, not necessarily mutually exclusive:
    1. Project was ill-planned.
    2. Migration to Linux / OO is not as easy as it seems.

    And, btw, it is not the file format of OO I am concerned with. I am much more interested in the support and development of software. I still don’t see a ready solid roadmap for LibreOffice. I don’t see one for OO, too.

  46. oiaohm

    Extremadura had an advantage they had not afford MS Office in the first place. So where using a different Office suit that was Linux friendly. So archives were not out to kill them.

    Also Extremadura was between a rock and a hardplace. They could not afford to pay MS for licenses yet need working and legal machines instead of a stack of illegal ones they had. Yes desperation equals fast risky migration.

    Yes Extremadura is a clear example why MS has so much trouble at times asking countries and companies to pay up. If you ask for more then they can pay they have no option bar to migrate. Not a migration I ever want to have todo.

    Basically its MS Office is normally the major vampire of migrations to watch out for.

    I have learnt check the MS Office usage before I attempt migration. Stack of Macros worry.

    Munich is very much the worse case. Long usage of MS Office lots of documents using Macros. If Munich can do it any business can do it. Just will take will. This is why so many people were 100 percent sure it would fail. The hardest part of the migration is done. That is MS Office to OpenOffice.

    If the migration to Openoffice failed the complete project would have been dead in the water. This is also why until the Microsoft Office to Openoffice issue was sorted out Linux machine migrations stopped. No point deploying Linux if you have to roll it backwards.

    It all comes down to your data how fast you can or cannot do migration.

  47. Robert Pogson

    Extremadura did 80K machines on the weekend. People came in to work on Monday with a new system.

    “Munich is receiving all the press about their careful and detailed migration to Linux on the desktop and here comes one of the poorest region in Europe showing that this can be simply done during a weekend.”


    It depends on breakage and manpower. I was breaking nothing and I extended my manpower by working 16h days and adding one guy for a couple of days for the really physical stuff like carting around thin clients. My employer went from 20 thick clients with no server to 153 thin clients and six servers and a gigabit/s network. I think the secretary kept her machine going with XP to avoid having to move her archives. Every other staff member went with Ubuntu GNU/Linux.

  48. Robert Pogson

    The migration was extended in years and increased in tasks. Those combined amount to a larger budget item but so would migrating to “7″ or replacing the servers or …

    Life goes on. Get over it. Munich has stated that giving more to the IT department has cost the taxpayers nothing extra because of the increased time and savings resulting from improving IT. Do you call it a disaster if you renew your mortgage?

  49. oiaohm

    Phenom did you not read the trend numbers I published since the middle of last year to the middle of this.

    >>June 2010: “More than 3000″
    >>February of 2011 more than 5000
    >>In June of 2011 more than 6500

    That is a 1500 per 6 months at worst. Yes they will hit the 12,000 the 80 percent requirement of the project on time in 2013 at current trend. At that rate total convertible machines should be done by the start of 2014. 2014 is when the interesting numbers will come out of Munich. The report here from robert says 10 percent not 20 percent windows. Yes the start of 2010 was a little slow. Last financial year trend is perfectly fine.

    Also to get the trend numbers you need to read German section of the LiMux site. The english site of LiMux is contains very badly out of date information with updated dates. Yes dates and information on the english section don’t line up. So giving the impression there is no progress. When in fact it is going like the bat out of hell. Yes the progress numbers in the Wikipedia for LiMux are correct if you dig in the German sections of LiMux.

    Reason for your progress number being wrong Phenom is that you have been dealing with people who have only read the english.

    Note all of 2010 the progress rate has been about 1500 thinking they did not start progressing on the Linux conversion to the end of 2009 until the OpenOffice conversion was done. Resources were fully focused on sorting that out.

    Phenom this shows how far behind you are. Munich is already using libreoffice in testing for the next version of there standard OS. OpenOffice and Libreoffice are Macro compatible. No extra conversion cost. In fact less conversion cost than converting between versions of MS Office.

    This is one thing that has been found in Munich. They have change versions of OpenOffice many times in the conversion timeframe. The bug bear has not been fixing up odf documents. The bug bear has been dealing with legacy MS Office stuff.

    Changing versions of OpenOffice caused less issues than changing versions of MS Office.

    Phenom its not escalating costs for Linux Migration for one simple reason. The conversion cost from Microsoft Office to OpenOffice is a one off expense. That expense ate the Linux Migration budget and time.

    Migrating between OS’s is expensive particularly for 12 000 of them. Unless those 12000 happen to be Linux. Like I have not had to format the machine I have been on for the Last 10 years and it still running the most current version of Linux.

    XP to Windows 7 you would be look at about 5.9 million for 12 000 machine migrated without licenses or hardware upgrades ie labor and training only. What is about 500 a seat. That is basically the normal rate.

    Remember due to Linux design this 5.9 million to convert the machines is basically a one off expense. Due to the fact Linux update system will keep all machines from this point on with the current OS. So OS conversion expense never happens again. Only costs then will be new machines.

    Staying on Windows and wanting everyone on the same OS will cost you that 5.9 million every 5 to 10 years with 15000 machines normally more with the the replacement hardware and the larger hardware windows requires.

    Phenom you really should have got a clearer picture where they spent extra.

    The one off expense on OpenOffice conversion will pay itself back due to deployment costs of MS Office in licensing alone savings.

    Yes you are free to claim changing from Microsoft Office to OpenOffice or Libreoffice can be down right expensive. But its about the same cost as staying with Microsoft Office.

    Phenom start reading the numbers and start blaming the right things. Most blame is OpenOffice at the time. Current day Libreoffice doing the same migration would be cheaper due to better macro support. Still going to be expensive.

    Linux conversion itself if you have no legacy dependency issues on Windows is not expensive.

    Basically watch out of the legacy because that is you budget doomsday.

    This is your problem Phenom big number panic something has to be wrong. Of course not being aware that 12000 machines is a darn huge number. Million across 10000 machines is only 100 to that machine. Serous-ally if you went out and acquire MS Office Enterprise licenses for 10 000 machines they are 500 a seat. There is 5 million. One MS Office upgrade cycle is 5 million without labor to install it.

    So total expenditure from 2005 to 2013. Will be recovered just in licensing costs saved by 2016 or before. If you take labor into account for deploying Microsoft software upgrades. Costs are currently equal. So first profits start of 2012. Yes all of 2012 and 2013 is bonus.

    This is why Munich is not particularly worried about the costs. They have not blown the budget as such. They have not come as far under as Munich had hoped for due to one off expenses. Really it was not realistic to aim for 50% cost saving straight up when you allow for the fact a deployment of this scale had not been attempted. 50% is what would have been achieved if the project completed on 2010. They are going to get about 25% straight up compared to a general MS deployment. That is including the MS Office to Openoffice migration that basic ate all the saving of that action in costs for that cycle with a complete date of 2013. The saving for that migration come in the next MS Office upgrade cycle when they don’t have todo that again.

    Basically the cost saving will keep on coming. Yes they could basically extend a year again and still be in cost savings.

    Phenom escalating costs claim requires you to know what the cost would have been if they had not started LiMux. Reason its not escalating is the fact is going to come in less than if they had stayed with Microsoft. Then on going cost saving are going to start of about 50 percent compared to if they had stayed with Microsoft. So leaving more budget for them todo other things.

  50. Robert Pogson and LibreOffice both use .odt files so are perfectly interchangeable with the documents. The feature sets of the two are rapidly diverging so it is unlikely they will ever merge and it is quite possible code will stop being shared because it will become too difficult to merge the changes.

    They are replacing 22 IT departments with one. Seriously, how do you think that will not save them money annually, forever? Then there are the constant upgrades to that other OS at great expense. That’s gone, forever.

  51. Robert Pogson

    Nonsense. By using FLOSS, my employer was able to buy more than a dozen printers, a bunch of cameras and scanners which they could not have afforded if they had bought thick clients with that other OS. We bought thin clients costing a bit over $100 each and build servers, saving $40K or so on software licences.

    File/auth server cost us $1800 and it would have been $1000 +$6000 for CALs more with that other OS. We actually built six servers and made 4 terminal servers. They would have needed CALs too so we saved $35000 just on server licences. Then 135 clients X $100 or so… We turned all that money that would have gone POOF! from the budget into valuable hardware providing even more functionality for the system. I was quite proud of the scanners. You could just plug one into any thin client in the building and use as a local scanner. Then we had 6 Xerox 8500 wax printers that were fast and gave brilliant colour. Kids and teacher loved them. With that other OS the school would have been stuck with sending kids to the office to pick up output.

  52. bilbophile

    Mr Pogson,

    My point was you did spent the IT budget during the migration although free software should have been free of cost and you did get paid for your work. Should your migrations have been budgeted as formal projects, these expenses would have been project costs. I presume these costs would have amounted to a relatively small but still significant share of the school budgets.

    Yet, you would have been paid the same wages even without the migration and the IT budget would still have been spent, although on software rather than on hardware.

    That is why – at least in your case – the migration was not a stress on your employer’s limited budget, despite the non-trivial total calculable cost of the project.

  53. bilbophile

    “Btw, they locked themselves with Open Office. Now, having the community flocked towards LibreOffice, I am curious to see how their Open Office is be supported.”

    This is irony, right?

    Seriously now, why do you think that according to the municipality (the link you provided) the increased cost of the project does not put additional stress on their total budget?

  54. FoSS Must D1e

    Posgon, it beats me how new 5.9 million euro is not escalating costs.

    When you are a FOSStard zealot it doesn’t matter how badly a project fails, it’s still a success.

    Pogson and Ohio Ham will always find a way in their brainwashed little minds to twist a story in such a way that it’s all of teh w1n for FOSStardia.

    We just laugh at them and realize that their miserable lives include dumpster diving for old computers and eating bulk mash potato mix.

  55. Phenom

    Posgon, it beats me how new 5.9 million euro is not escalating costs. They still have converted less than half of all workstations, and if they keep the current trend (last one year), they will not make it in 2013.

    Btw, they locked themselves with Open Office. Now, having the community flocked towards LibreOffice, I am curious to see how their Open Office is be supported. I guess they will need to wait for Apache to get their act together, before they can be sure they have a solid ground.

    Nice job, indeed. In corporate environments, that would be certain managers getting the boot.

    In Munich, it is likely that a certain party is getting the boot on the next elections.

  56. oiaohm

    Robert Pogson Munich have the issue of being active in use and migration has to happen as well.

    Yes 10 day migration is workable in 153 client machines.

    Rapid migrations fall apart at about 200 machines. Reason support people normally drown. Too many calls too quickly. Its bit like a swimming pool. At some point you exceed the life guards.

    Yes funny enough the capping limit is normally the telephone system.

    40 machine blocks as I do. Normally is small enough that there is not a big shell shock problem. So mostly number of errors coming out of them is close to normal network errors.

  57. Robert Pogson

    I have a different style than Munich. I throw people in and help the ones who cannot swim. No one drowned yet…

    I once planned how to convert a lab in about two hours, one to install and configure the terminal server and the other to get the PCs to boot PXE to become thin clients. I did set up an entire school with 700 accounts and 153 clients in ten days. I oriented the teachers and gave them lists of accounts to distribute to students. The plan was for a one-month migration but the building was new and the handover did not happen until late in August. It was hot too… Cost of labour per PC was about $35. At that rate, the Munich migration would take me 1000 days and cost $490000 for labour. No doubt that would be a disaster/riot.

  58. oiaohm

    bilbophile Yep 100 percent right.

    Biggest issue here is people seeing big number and attempt to use those to create a panic instead of doing a proper assessments of where that money is going and what is going on.

    Results with LiMux look very sane. Important lessons but sane.

  59. bilbophile

    If the release is accurate, considering LiMux a failure is like saying that Mr Pogson’s Linux conversions were disasters because they took a whole year to complete and were too expensive because they included Mr Pogson’s wages for one year plus all the budget he spent on hardware upgrades rather than on software licences.

  60. bilbophile

    I don’t read German and your first link does not work anyway. However the Google translation of the relevant page seems to be a little more nuanced.

    “The General Assembly of the Munich City Council decided to address in IT project LiMux the IT process professional[ism], a higher level of IT standardization and other technical improvements for the city administration office jobs. The management expects that this will permanently lower IT operating costs, as well as more acceptance for the single office workstation and implements technical requirements that have changed significantly since the project began.

    The IT project for LiMux is extended until 2013 and the project budget [is] increased by [an] estimated additional expense of € 5.9 million.

    The added expense comes in the years 2012 and 2013 and does not lead to increased stress on the city budget, because all departments and municipal enterprises and to finance the expanded approach from existing budgets.”

    The way I read this, the Council approved an upgrade of the project objectives in line with the evolution of the IT industry since the original approval of the project.

    According to the Council the additional budget will not be reflected in an increase of the general municipal budget. It is therefore likely that the added costs consist in wages of the municipal IT staff – who will be working two extra years for this project -, hardware replacements that should have occurred anyway but,as a result of the deadline-extension, will become part of the project, other concurrent or subsequent IT projects which had to enter under the umbrella of the LiMux project as a result of its extension/expansion.

  61. oiaohm

    “1500 machines in under 5 months. So that puts it on time frame to hit 10 000 by the mid of 2012. With a possible hitting 12 000 by the start of 2013. If it does not slowdown complete take over by end of 2013 at latest middle of 2013.”

    Typo. Last 2013 should be 2014.

  62. oiaohm

    Phenom of course you like to miss context.

    5.9 million euro is a 24/7 support contract level support including custom application development and maintenance. Phenom. That is a 3 year extention. So about 2 million a year.

    You should try getting support for windows sometime for 6900 PCs at 24/7. Licensing is only first half of the battle. Its not cheap providing proper support to back it up.

    Once the number crosses 10 000 machines it will not be looking to bad at all. Anti-virus cost is about 50 dollars per machine per year for Windows. So that is 1/4 of a million at 6900 a year. Changing to 1/2 a million at 10 000 machine.

    Its very simple to forget all the hangers on Phenom when using Windows. Including the developers making your internal applications.

    Also Phenom something you have missed Munich is already 100 percent migrated to openoffice/libreoffice. So returning to Windows does not require Microsoft Office licenses any more. Everything that depended on Microsoft Office is already gone. Also the openoffice migration was the expensive bit.

    Also everyone is using the web-services and other remote access services to increase productivity. Items that otherwise would cost.

    Phemon from the outset Munich never said 100 percent migration would ever happen.

    Really cost of migration has not be any more than cost of support that would have had to been paid anyhow. With no possibility of expanded function or cost reductions.

    Phemon this is a key point.
    “Heavy use will be made of web-applications to facilitate future changes to the system.”
    Result less of the operations are OS dependent. Less data spreed all over the place to backup as well.

    There is more than cost saving here. There are data integrity changes as well.

    LiMux is not a desktop only project its server room as well. Do you think maintaining a government websites are cheep?

    Also is moving at a quite decent pace these days on migration as well
    >>November 2008: 1200
    >>June 2010: “More than 3000″
    >>February of 2011 more than 5000
    >>In June of 2011 more than 6500

    1500 machines in under 5 months. So that puts it on time frame to hit 10 000 by the mid of 2012. With a possible hitting 12 000 by the start of 2013. If it does not slowdown complete take over by end of 2013 at latest middle of 2013.

    Also interesting you claim no acceleration. 2009-2010 number of machines migrated is now being done in 5 to 6 months. I would call this acceleration. Quite impressive acceleration.

    Of course the 12 000 in sometime in 2012 is quite a reasonable goal. So really it don’t need to accelerate any more Phemon its most likely moving fast enough to hit current set goal with current move speed 1500 every 5 to 6 months.

    Now lets do some maths here how many hours per machine. There is 5 days a week 8 hour days times a 14 weeks. 560 hours. number of staff 50, 28 000 divide by number of machines 1500 gives approx 18 hours per machine or just over 2 days per machine for training and issue sorting. That is allowing for 1500 over 6 months not 5.

    2 days per machine is also about the average for deploying Windows 7 as well. So for a complete change of OS I think its quite impressive that the number is about the average cost in time as people upgrading windows.

    Yes 2010 they revised to a more practical number for 50 people to migrate and fix any issues found. Just work out the wages for 50 support personal Phemon. That alone is not cheep.

    Of course limuxwatch has not been charting progress to see that the claimed figures are 100 percent possible. Yes the claimed figure for 2013 is possible.

    Now if you had charted the goal of 100 percent complete by 2010 with number of man hours it was kinda insane it was less than 1 hours per machine due to other failures compressing the time to migrate to nothing. So of course that goal failed.

    Reason why the hours for the migration to Linux where not there for 2010 is that the migration to OpenOffice was completed 3 years behind schedule due macros hit that were not portable. And the start of the OpenOffice migration was 1 year behind schedule came from the trial period being extended by extra 12 months and the end date for the project not be extended accordingly.
    “September 6, 2005 It is decided that the project needs an additional one year pilot test, and migration slips one year.”
    At this point the project could have still been given up.

    Basically the migration to Linux has been running as per clockwork. Phenom. It was the migrations of applications on Windows that blew the budget.

    So over all its only 3 year outside forecast. Really only 2 when you allow for the fact the end date was not extended when the test pilot stage was extended. This is a major project management error really. You cannot compress time no matter how much you want.

    Even better the 2 years out is fully accounted for. Nothing todo with Linux on desktop. Everything todo with migration to OpenOffice. Don’t take it cheep.

    Now does not limuxwatch statements look like a raving fool.

    Lesson to take away from Munich don’t worry about Linux Desktop its not the thing that is going to head hunt you. Worry about your applications produced data. That is what is going to bite you. Migrating on Windows or Linux the Applications can be out to eat your time alive.

    Yes shock of shock. Migrating to Linux is not a bad as people make out if Munich makes it on time for 2013. Everything is possible at this stage.

  63. Robert Pogson

    As I wrote and the facts support it rather than just migrating what they had to GNU/Linux, Munich is also completely reorganizing their IT. That is what the extra money is for. They still have not used all their original budget for the migration. 5.9 million Euros over years is small change for an organization that size. One round of M$’s licensing is many times as much and Munich will be saving that every few years forever.

    The costs are not escalating. They are doing it largely with existing staff and probably actually reducing staff because they are rolling 22 IT departments into 1. Munich has responsible government. If it were a disaster the tax payers would have revolted.

  64. Phenom

    Pogson, this is outright laughable.

    You convenienly miss one major point of your news:
    “The IT project for LiMux is extended until 2013 and the project budget to the estimated additional expenditure of € 5.9 million increases”

    5.9 million euro! Just to the math yourself how many Windows and Office subscriptions (not retail, mind you!) that would buy, including servers, and reduce time because there will be no need to migrate existing applications, documents, and to educate the workers anew, thus inevitably lowering their productivity for quite some months at least.

    If you would take your time to read carefully, you will see the Gargantuan scope of FAIL that Limux is.

    The key point – the goal post for completing migration is constantly moved forward, the pace is never accelerating, while the costs of migration go higher and higher.

    Sane people from the IT business call this disaster.

Leave a comment