> Today the CSU political party, which has a long track record of opposition to LiMux, is also part of the ruling coalition in Munich. It was this coalition of CSU and SPD politicians that put forward the proposals to switch back to Windows 10 earlier this year.
The reality is enterprise support for linux systems is the one variable i don't really see discussed here. This is Microsoft's bread and butter for a reason. Organizations prefer an easy upfront cost rather than complex, support intensive solutions.
It depends on the type of support. I'd expect a place such as the city hall to be somewhat low tech, regarding user sophistication levels. And in such an environment most of the support requests are the "where did the button go?" type.
A switch to Linux has a good chance to increase these types of requests, at least in the near term.
Users should not be fiddling with the parts of the environment that change between Windows and Linux. Users should not be sticking USB sticks, setting up printers and so on. They should run the programs they are supposed to run in order to work.
Maybe they have to touch those things in order to make their workflow optimal, or even work in the first place.
One of my big issues with the Linux Desktop community is this attitude that they understand your workflow better than you do. You ask how to do something and you get variations of "don't do that".
Except that, again, you're making assumptions. You're assuming you know why they are trying to do what they're asking about, that they haven't rejected the 'more appropriate' solutions for reasons that are quite possibly outside their control. I have been in this situation a fair amount and it is like pulling teeth to get some people to answer a simple question.
I think that the main problem when it comes to Linux support is that people who really understand Linux/UNIX have an easier time finding better paying positions. Lower paid IT workers tend to hang around the MS ecosystem. This shows up in the cost of maintenance for Linux solutions.
I think so. Hiring half decent Linux support staff would be a nightmare.
Window support staff are cheap and abundant. Not to mention the off the shelf outsourcing providers. IT in most large organizations already happens in India.
Love Linux and use it as my main desktop. Supporting corporate wide? Big no.
It's funny, I read: "AFAIK Accenture engineers were handling linux administration of Munich municipality" as confirmation that hiring half-competent team of Linux admins is difficult. But I confess to being biased to the corners of Accenture that I've been previously exposed to...
Linux is a lot harder to maintain and it's usually for servers (read: important), companies pays for the experience. Any dude who can administer Linux should rebrand himself to devops and double his rate.
Windows is easier to maintain and more intuitive. There is always a dude who can make it work with a bit of point and click.
People who can administer both will eventually end up on a Linux jobs because they pay more and they don't risk being the monkey who maintains the printer.
The article didn't mention what the yearly Windows licensing costs would be after the transition. I remember that the elimination of such costs was an argument for the transition in the first place.
I'm sure Munich council's decision to switch back to Windows, despite all the costs and long-term lock-in, had nothing to do with this totally random HQ-building in Munich:
If MS built their HQ in Munich because of that then it's an absolutely amazing deal for Munich.
Just the tax on salaries from those 1900 people working there will pay off the costs of switching to Windows in a very short time.
And as a nice bonus I'm sure that 28% of users that "have severe issues related to software, which could be solved by migrating these users to Windows and MS Office."[0] will also be happy.
They moved their HQ from just outside the city border to inside the city. I doubt many of the employees will move (salary is taxed at the place of residence).
I wonder how they plan to disable the spyware? Even the enterprise edition doesn't allow turning everything off.
My employer is still rocking Windows 7 for this reason.
At one point we trialed Windows Phone, the guide was a series of embarassing turn this feature off, turn that frature off to try to limit the amount of data being syphoned. Now we use iPhones.
Ostensibly. Given the nigh-malware-like tactics that W10 uses to work around those who try to disable some of that behavior, I'd be quite surprised if it was completely clean.
And if that fails, it's till s a massive blob of closed source proprietary code at the end of the day. Your single option is to trust Microsoft.
> I'd be quite surprised if it was completely clean.
I understood that LTSB was intended specifically for semi-critical/continous usage applications, like airport kiosks and the like.
To me, it would seem a bit irresposible to pollute LTSB with telemetry and cruft, when Microsoft knows full well that businesses bought that distribution for that specific guarantee.
But micrisoft doesnt wants general PC's running LTSB. As they put it ..
"Windows 10 Enterprise LTSB is a separate Long Term Servicing Channel version. Long-term Servicing channel is not intended for deployment on most or all the PCs in an organization; it should be used only for special-purpose devices. As a general guideline, a PC with Microsoft Office installed is a general-purpose device, typically used by an information worker, and therefore it is better suited for the Semi-Annual servicing channel."
That's not at all what it says. The cost of moving to Windows and office is ~€5 million:
>€ 4.8 million and licenses (for Microsoft Windows, Microsoft Office, software distribution,
The €50 is for hardware replacements that were going to happen anyway.
Nothing in the article addresses the harder to track costs of user efficiency and expert support staff. I'm not claiming it's cheaper in either direction, but staffing costs are a very, very real cost.
> the former open-source champion would pay €49.3 million to create a new Windows 10 client OS
I don't know what you mean by calling a "new client OS" "hardware replacements that were going to happen anyway".
I find the sentence confusing itself, as the OS would be microsoft, maybe they mean a distribution with preinstalled software? Either way, I don't read it to mean hardware replacemens
I find it equally hard to read that sentence and assume they're actually spending €50m to create a new OS when whole point of the article is that they're switching to an OS that already exists. the only assumption i can make is that the author of the article doesn't know what they're talking about.
> The €50 is for hardware replacements that were going to happen anyway.
What do you think why people complain about Limux all the time? Hint: the PCs, keyboard, mice, screens - everything WAY outdated. Most, I'd take a guess here though, not even able to run W7.
Also, LOL 4.8 M€ only for total cost of licenses and migration? Count that x4, at the very least.
Shifting the blame on Limux instead of the lack of money for proper hardware is the easy way of justifying to shift millions upon millions of dollars into the coffers of Accenture and Microsoft.
I don't know how limux was done, but modern hardware may give just about nothing to the users.
Office was single threaded until 2007, companies with previous office licence wouldn't upgrade, but they had modern thinkcenter desktop with good dual core cpus yet their excel spreadsheet would choke on a single core all day long.
A good information system is almost an art, it requires proper knowledge from hardware, network config (<= most probable performance bottleneck for the users), software and end users.
Nobody does this because it's too subtle and not rewarding enough. But if someone did, any city could operate on "old" hardware using linux. I don't speak for the office suite because LibreOffice [1] is not good enough IMO. But for the rest it could be even better than recent win10 machines.
[1] I love their work, it's amazing but.. still not there yet, just like firefox and chrome.
> I don't know how limux was done, but modern hardware may give just about nothing to the users.
I am talking about machines with less than 2GB of RAM, running modern versions of Firefox, Thunderbird and OpenOffice in addition to all the custom software required for the jobs (including, for some, the Java SAP client). You get why people complain about the computers being slow? FF is notorious for memory hogging, Thunderbird is not much behind it, and OpenOffice and SAP GUI are Java crap which means a tendency to leak memory the longer you have it running and general downsides of simply being Java. The machines swapped like hell, and SATA disks were not a given.
In addition, web sites themselves have grown in resource usage in an exponential way. HN and Fefes Blog are the only two websites I know that I can use even when my data allowance on my phone is finished and I'm throttled to 56k. Everything other? Forget it. And it's not just data usage, it's also RAM and CPU usage, e.g. due to ads and dozens of trackers.
> network config
Actually, it's not the config that's the bottleneck - it's the uplinks and crosslinks between the dozens of city buildings which includes the schools - mine, for example, was linked to the central DC with a 10/10 fiber 14 years ago and still was when I left the school 6 years ago, you can guess how much "fun" that is to use...
But the resource sinks are also something absurd that should be avoided. I've seen this first hand. Lots of business were seduced into migrating onto web based application. I was crying ..
Sure the old terminal apps were crude bug god they were fast, processing AND ergonomy. The new one looked fancy but wasn't a production tool. Came late and under delivered. And did less than the old tools by design.
It's all bs spread by consultants to gain market shares. Employees and consumers don't need this. The energy is spent where it shouldn't at the benefit of the wrong persons.
> Sure the old terminal apps were crude bug god they were fast, processing AND ergonomy.
The old terminal apps are mostly designed to run in batches, and it's horrifically difficult to implement (secure!) interfaces to any modern systems - which is expected from a government these days. There's a reason why many banks still don't offer instant payments in 2017 - not just because a tiny profit can be made by holding the money for a day, but also because the mainframe based batch processing stuff is not able to do realtime.
And to make stuff worse: many of these terminal apps have severe issues. They don't accept UTF8 (some only ANSI!), leading to data loss when you have customers/citizens with uncommon characters in their names - while e.g. a German system will often accept the Umlauts, it will struggle with French or Slavic diacritics and completely fail at Cyrillic or Japanese. Or they're hardcoded for example to encode gender as bool (0=male, 1=female), which breaks with intersexual people, or that a marriage may be between homosexual people (because the data fields are named as husband / wife)...
Sure, the same issues plague web-apps or native desktop apps too, but for these you can at least hire decent programmers pretty much anywhere while you will spend enormous amounts of money for mainframe experts.
- 14 Mio € for personal costs
- 24 Mio € for consulting (not sure why that is that high?!)
- 13,4 Mio € for it@M services (support, hardware installation, etc it@M is the it company that manages the project)
- 4,8 Mio € for HARDWARE
- 29,9 Mio € for Licenses (Microsoft Windows, Microsoft Office, software delivery, License-, Print-, and Profile Management, identity management and extension of the virtualisation environment)
(to be fair the pdf quotes that a big cost will be the investment into additional virtualisation, probably licenses)
I've worked in industry long enough to suspect that there's an excellent chance that any quoted numbers are so adjusted by whoever is most politically powerful that my intuition is to just shrug my shoulders and move on to the next article. Dishonesty is so prevalent in so many "news" articles today a person should just read things and have a little chuckle before going about their day.
The hardware replacement cost must be realised because of Windows 10 compatibility, the old computers don't support it. Same problem hits on Win 7 to Win 10 transition.
They should just get office 365 and use the online version of word Excel and PowerPoint in chromium. That's better than libre office and they won't need to switch to windows 10.
The proprietary nature of the software doesn’t matter. All that matters is that their FILES can be read and written. And with Office XML, since 2007, this is a solved problem.
IIRC, it is essentially impossible to correctly implement parts of OOXML without reverse-engineering the Office suite. The standard does not provide enough information.
Online versions still pale in comparison to the desktop client. I'm sure people in accounting would be upset by a forced transition from desktop Excel.
To be fair accounting is supposed to use specialized software, not excel to do the actual accounting. Software that could absolutely be web based.
The problem I suppose is more hardware issues like printers, card readers and a lot of third party professional software that just doesn't have drivers for linux.
I think Microsoft must have a very powerful lobby tackling this potential "end of an era" event.
With more companies moving analytics from Excel to Python, other data science tools, cloud or not... And governments adopting ODF as document standards.
Not to mention the free beer.
Python works for data scientists, but the people who use Excel are not all data scientists. Python and Excel are as comparable as apples and oranges. They're both fruit, but that's literally all they have in common.
People are looking for something better than Excel though, particularly for teams. Raw python won't work, it would be a chaos of copy pasted receipes and voodoo... VBA is bad enough.
The spreadsheet model could be leveraged, but without free form tables and formula repetition. I'm torn about whether allowing explicit recursion or iteration is a good idea. It would have to include a visionary visual debugger.
Yes. But one of the arguments, in my previous company, for Excel over LO Calc, was speed for lots of data, data sources speed... With more tooling around the end user data will be less and more clean/prepared.
Moving excel to python is entirely irrelevant here because this subject is not about an average human trying to use a text editor for general things (like writing documentation).
That's how it works in Desktop Linux. The system is such a hodgepodge of interconnected pieces, each with their own slight variations in operation, configuration, and quirks, that once you finally put something together that works to your tastes you pretty much have to create your own distribution to keep it that way.
...or you can just contract Red Hat or SUSE (which used to be a German company) to standardize on RHEL/SLED plus some customizations. They would be a pretty big customer with a good perspective to add other municipalities and branches of government in the case of success.
Rolling your own distro is a sensible thing to do for a big installation. Thanks to open source, you can take an existing distro (Ubuntu, in this case) and modify it, until it suits your needs. Some people prefer to call such a thing »spin« or »blend«.
That is what they're wanting to move to. They asked the Microsoft partner Accenture, who could sell a lot of Windows clients here, to evaluate their IT situation and not even they dared to throw their reputation away and somehow suggest that Windows is a necessity. They also just suggested a thin-client architecture.
> ...will also reduce costs by not having to run a Windows and LiMux client side-by-side.
Mind that this means "reduce costs in comparison to moving to another non-Windows OS". They still do not have a cost calculation that would promise long-term savings and I'd call it unlikely at least that there will be savings, as the previous city mayor reported huge savings after the migration to LiMux.
77 comments
[ 3.8 ms ] story [ 133 ms ] threadHow do you say "pork" in German?
The reality is enterprise support for linux systems is the one variable i don't really see discussed here. This is Microsoft's bread and butter for a reason. Organizations prefer an easy upfront cost rather than complex, support intensive solutions.
A switch to Linux has a good chance to increase these types of requests, at least in the near term.
One of my big issues with the Linux Desktop community is this attitude that they understand your workflow better than you do. You ask how to do something and you get variations of "don't do that".
> You ask how to do something and you get variations of "don't do that".
I usually add "do [something appropriate] instead, which will solve the problem and not introduce a [horrific IT risk]".
Window support staff are cheap and abundant. Not to mention the off the shelf outsourcing providers. IT in most large organizations already happens in India.
Love Linux and use it as my main desktop. Supporting corporate wide? Big no.
Windows is easier to maintain and more intuitive. There is always a dude who can make it work with a bit of point and click.
People who can administer both will eventually end up on a Linux jobs because they pay more and they don't risk being the monkey who maintains the printer.
I imagine this is true for a lot of businesses as well.
Nothing against, Linux for the desktop but enterprise wide is a big difference.
https://www.itworld.com/article/2716115/operating-systems/sw...
I'm sure Munich council's decision to switch back to Windows, despite all the costs and long-term lock-in, had nothing to do with this totally random HQ-building in Munich:
https://mspoweruser.com/microsoft-germany-moves-into-a-new-h...
Just the tax on salaries from those 1900 people working there will pay off the costs of switching to Windows in a very short time.
And as a nice bonus I'm sure that 28% of users that "have severe issues related to software, which could be solved by migrating these users to Windows and MS Office."[0] will also be happy.
[0]: https://blog.documentfoundation.org/blog/2017/02/14/statemen...
My employer is still rocking Windows 7 for this reason.
At one point we trialed Windows Phone, the guide was a series of embarassing turn this feature off, turn that frature off to try to limit the amount of data being syphoned. Now we use iPhones.
And if that fails, it's till s a massive blob of closed source proprietary code at the end of the day. Your single option is to trust Microsoft.
I understood that LTSB was intended specifically for semi-critical/continous usage applications, like airport kiosks and the like.
To me, it would seem a bit irresposible to pollute LTSB with telemetry and cruft, when Microsoft knows full well that businesses bought that distribution for that specific guarantee.
"Windows 10 Enterprise LTSB is a separate Long Term Servicing Channel version. Long-term Servicing channel is not intended for deployment on most or all the PCs in an organization; it should be used only for special-purpose devices. As a general guideline, a PC with Microsoft Office installed is a general-purpose device, typically used by an information worker, and therefore it is better suited for the Semi-Annual servicing channel."
from: https://docs.microsoft.com/en-us/windows/deployment/update/w...
>€ 4.8 million and licenses (for Microsoft Windows, Microsoft Office, software distribution,
The €50 is for hardware replacements that were going to happen anyway.
Nothing in the article addresses the harder to track costs of user efficiency and expert support staff. I'm not claiming it's cheaper in either direction, but staffing costs are a very, very real cost.
Isn't Windows is the reason why they want to tranformation from Linux to Windows?
The original documents say something else. The original document mentions 89 million EUR. If the translation is half correct:
https://translate.googleusercontent.com/translate_c?act=url&...
https://translate.google.com/translate?sl=auto&tl=en&js=y&pr...
> the former open-source champion would pay €49.3 million to create a new Windows 10 client OS
I don't know what you mean by calling a "new client OS" "hardware replacements that were going to happen anyway".
I find the sentence confusing itself, as the OS would be microsoft, maybe they mean a distribution with preinstalled software? Either way, I don't read it to mean hardware replacemens
What do you think why people complain about Limux all the time? Hint: the PCs, keyboard, mice, screens - everything WAY outdated. Most, I'd take a guess here though, not even able to run W7.
Also, LOL 4.8 M€ only for total cost of licenses and migration? Count that x4, at the very least.
Shifting the blame on Limux instead of the lack of money for proper hardware is the easy way of justifying to shift millions upon millions of dollars into the coffers of Accenture and Microsoft.
Office was single threaded until 2007, companies with previous office licence wouldn't upgrade, but they had modern thinkcenter desktop with good dual core cpus yet their excel spreadsheet would choke on a single core all day long.
A good information system is almost an art, it requires proper knowledge from hardware, network config (<= most probable performance bottleneck for the users), software and end users.
Nobody does this because it's too subtle and not rewarding enough. But if someone did, any city could operate on "old" hardware using linux. I don't speak for the office suite because LibreOffice [1] is not good enough IMO. But for the rest it could be even better than recent win10 machines.
[1] I love their work, it's amazing but.. still not there yet, just like firefox and chrome.
I am talking about machines with less than 2GB of RAM, running modern versions of Firefox, Thunderbird and OpenOffice in addition to all the custom software required for the jobs (including, for some, the Java SAP client). You get why people complain about the computers being slow? FF is notorious for memory hogging, Thunderbird is not much behind it, and OpenOffice and SAP GUI are Java crap which means a tendency to leak memory the longer you have it running and general downsides of simply being Java. The machines swapped like hell, and SATA disks were not a given.
In addition, web sites themselves have grown in resource usage in an exponential way. HN and Fefes Blog are the only two websites I know that I can use even when my data allowance on my phone is finished and I'm throttled to 56k. Everything other? Forget it. And it's not just data usage, it's also RAM and CPU usage, e.g. due to ads and dozens of trackers.
> network config
Actually, it's not the config that's the bottleneck - it's the uplinks and crosslinks between the dozens of city buildings which includes the schools - mine, for example, was linked to the central DC with a 10/10 fiber 14 years ago and still was when I left the school 6 years ago, you can guess how much "fun" that is to use...
But the resource sinks are also something absurd that should be avoided. I've seen this first hand. Lots of business were seduced into migrating onto web based application. I was crying ..
Sure the old terminal apps were crude bug god they were fast, processing AND ergonomy. The new one looked fancy but wasn't a production tool. Came late and under delivered. And did less than the old tools by design.
It's all bs spread by consultants to gain market shares. Employees and consumers don't need this. The energy is spent where it shouldn't at the benefit of the wrong persons.
The old terminal apps are mostly designed to run in batches, and it's horrifically difficult to implement (secure!) interfaces to any modern systems - which is expected from a government these days. There's a reason why many banks still don't offer instant payments in 2017 - not just because a tiny profit can be made by holding the money for a day, but also because the mainframe based batch processing stuff is not able to do realtime.
And to make stuff worse: many of these terminal apps have severe issues. They don't accept UTF8 (some only ANSI!), leading to data loss when you have customers/citizens with uncommon characters in their names - while e.g. a German system will often accept the Umlauts, it will struggle with French or Slavic diacritics and completely fail at Cyrillic or Japanese. Or they're hardcoded for example to encode gender as bool (0=male, 1=female), which breaks with intersexual people, or that a marriage may be between homosexual people (because the data fields are named as husband / wife)...
Sure, the same issues plague web-apps or native desktop apps too, but for these you can at least hire decent programmers pretty much anywhere while you will spend enormous amounts of money for mainframe experts.
Source: https://www.ris-muenchen.de/RII/RII/DOK/SITZUNGSVORLAGE/4740...
Somebody read over the 'and' (edit:) and to be fair it's a wierd way of expressing something that is so important in german. probably an intention.
A government should not set itself up in a way that requires software from a specific vendor to interact with it.
[1] https://en.wikipedia.org/wiki/Office_Open_XML
The problem I suppose is more hardware issues like printers, card readers and a lot of third party professional software that just doesn't have drivers for linux.
With more companies moving analytics from Excel to Python, other data science tools, cloud or not... And governments adopting ODF as document standards. Not to mention the free beer.
* Governments/Councils start using this as an example.
* Companies that work with the council start supporting Linux and LibreOffice.
* Council worker might get a Linux as the next personal laptop buy.
The spreadsheet model could be leveraged, but without free form tables and formula repetition. I'm torn about whether allowing explicit recursion or iteration is a good idea. It would have to include a visionary visual debugger.
Rolling their own distro was a terrible idea.
Mind that this means "reduce costs in comparison to moving to another non-Windows OS". They still do not have a cost calculation that would promise long-term savings and I'd call it unlikely at least that there will be savings, as the previous city mayor reported huge savings after the migration to LiMux.
Yikes!
Wasn't the the migration away from Windows something like €30m?
OTOH, has nobody written an Atom-based Windows emulator yet that runs in a Docker container?