Hacker News Reader

Comment by wyager | Hacker News Reader

What are the reasons for the government not to switch to linux? I'm not trying to be snarky; it really seems like there are a huge number of benefits and very few downsides.

I understand that the government has a contract with MS, but it seems like it might actually be costing the government more to take advantage of the contract.

View on HN

At their scale, they will need to pay someone for support no matter what, whether that "someone" is Microsoft, Red Hat, Canonical, etc.

So, you can't just look at the consumer price tag on a Linux distribution ($0) and compare that to what the IRS is paying Microsoft in support contracts - you have to compare it to the total costs of enterprise Linux support.

And if the IRS is having a hard time migrating off of a version of an OS that was released 12+ years ago to its more recent version, I imagine they'd have a hard time migrating to a completely different operating system altogether.

Even if the incremental costs of upgrading LTS versions of a Linux distro ended up being smaller, they simply don't have the up-front capital to cover the costs of migrating to Linux in the first place.

(Oh, and don't forget that even if they did manage to procure that somehow, as soon as Microsoft gets wind of it, the price for Windows support contracts will magically fall).

To be honest, I don't see how migrating to Linux would actually solve the problem the IRS is having. If anything, it would make the problem worse. Ubuntu LTS is only supported for 5 years on the desktop. RHEL goes up to 13 years, which is about how old Windows XP is right now. Windows or Linux, the IRS would have run into migration pain as support for its outdated operating system ended. Yes, if they'd used Linux, they would have had the possibility of backporting fixes. Realistically, though, very few organizations outside of the distros themselves backport kernel fixes. Instead of hearing about the IRS making extra payments to Microsoft, we'd be hearing about them making extra payments to Canonical or Red Hat.

Now I'm not saying the move to Linux is necessarily the right move for the IRS, but you missed the point entirely. The only thing stopping the transition on time was that the upgrades were put on hold because of budget issues, not because it was hard to upgrade. Last I heard, Ubuntu was free to install, so sure they would still have a similar upgrade schedule, but this wouldn't be an issue here, at least in terms of budget for such an upgrade.

You're just thinking about the cost of the OS licenses, think about how many man hours it would take to update all those machines.

Oh come on, the man hour cost to install the update is miniscule to the cost of a Windows license per machine (even if they do get site licenses). And if their problem is budget, switching to Linux is a potentially valid fix for this problem, as the cost (even though there is still a cost) would be dramatically lower. Although again I'm not advocating for the change as there are many other issues that would be caused by this transition.

Most places I've worked - organizations far smaller than the IRS - have had site licenses, so the costs of upgrading don't lie in paying for the OS.

The costs lie in training, upgrading hardware, and upgrading user applications (including things that are closed source, discontinued or developed in house by people who have left the organization).

Unfortunately you'd get most of these problems in Linux too.

If you read the article, they have already upgraded some of their computers, so the upgrading user applications has already been done, as the machines that have been upgraded would have needed this. The training required to go from one version of an OS to another version of the same OS, is very miniscule, as (at least for the standard user usage, which the majority of IRS users, they fall under) OS developers go out of their way to make the changes as seemless as possible. And sure upgrading hardware may be a cost, but again this generally a much smaller problem with Linux, as Linux is generally a much lighter weight OS than Windows is.

> If you read the article, they have already upgraded some of their computers, so the upgrading user applications has already been done, as the machines that have been upgraded would have needed this.

But they've upgraded to another Windows version, not a * nix. Windows has taken backwards compatibility support to plaid (11 if you're a fan of a different movie). I worked in an office with a data analysis tool written using Visual Fortran in the 1990s for Windows 95. Without retargeting it it runs just fine with a single launch flag set for Windows to use the appropriate compatibility mode. The cost of upgrading user applications is really the cost of testing user applications. Moving to * nix wouldn't be upgrading, it'd be rewriting/porting or significant testing with Wine or some other compatibility layer. [1]

[1] As others pointed out, as more applications run on the CLR porting to * nix will become easier as CLR support improves on those platforms. But that also requires a rewrite, you can't take an existing Fortran or VB or C++ project and retarget it without effort.

The issue is clearly not about the price of the licenses. The IRS is paying 200$ per machine to extend XP support (which is more in a single year than what they would have paid for new Win7 licenses).

The expensive part is upgrading the hardware to support Win7, rewriting all the software that aren't compatible with Win7, training the personnel to use Win7 (and all the newly re-written software). Now, imagine if they were actually trying to upgrade to Linux. They'd pay much more to train the personnel, they'd have to re-write almost every single software they have been using. On top of that, they'd still pay for custom support (and Linux support is considerably more expensive than Windows support).

>Linux support is considerably more expensive than Windows support

In my experience running a small business, Linux support was much cheaper than windows support. The Linux professionals seem to charge more but can almost always fix a problem twice as fast as a the support I get from Windows "professionals". I also ended up having less problems with Linux than I did with Windows (after a few initial bumps in the road). I've also talked to quite a few other entrepreneurs and many of them agree that Linux is a fine and much better choice for start-ups simply because you're starting from a clean slate (So there is no legacy code to deal with) and because 90% of small businesses and start-ups consists of a network of 10 or less computers.

This might be true for startups, but the IRS certainly have more than 10 computers.

>Last I heard, Ubuntu was free to install

Ubuntu is free to install only because your free time has no monetary value. For organizations whose time has monetary value, the cost of an OS license is swamped by the cost of the time it takes to rewrite applications, retrain users, roll out the upgrade, and fix the inevitable breakages.

And like I said above, these costs are present whether you're using Linux or Windows.

Also, if you are the type of person who hangs around on HN, you are more likely to be the type of person with friends who run linux or who is comfortable working through software configuration problems by googling for things and trying thinks from internet forums. Most people don't have that privilege, and so linux is actually pretty expensive for them in terms of time and frustration.

(I run ubuntu with xmonad but have never written my own packages)

I think it would be great for them to switch but I can think of 2 huge reasons they wouldn't.

1. Retraining the users, very few people use linux so it would be a massive undertaking to retrain their thousands of mostly non-technical workers to use a new, unfamiliar, operating systems.

2. Microsoft Office. Say what you want about LibreOffice and OpenOffice but when it comes to enterprise grade software MS Office is unrivalved, especially concerning Excel.

3. And whatever other IRS-focused enterprise software they may be running, probably runs on the Windows platform.

Side note: Maybe this gets better in 10 years when lots of things have been written on the managed CLR? Although I don't think the challenge Wine has is running the C++ programs, but implementing all of the stubbed out APIs correctly... the same challenge exists with a C# interface.

4. As good as Ubuntu's latest version and its UI may be right now, anything other than basic stuff you want to do requires command prompt. Normal users who are not geeks can't imagine themselves doing that on a daily basis.

Still wondering how they even go by doing their job on a computer. If you ever worked in a support role, you'd know that no matter how many years of experience they have, they are most likely being confused by the UI change every time MSFT decide to be 'cool' and 'innovate' (See the ribbon style introduction back in 2007). And this is without the regular nagging of them not wanting to learn where are the options they want to modify, despite the menus and sub menus.

With regards to MS Office and MS mail client, sure they are more polished, but does that really matter? Are those people crafting an art piece every time they want to add an image to their report? Just an honest question.

3. Microsoft Outlook. If you've worked at any of these big organisations, you quickly learn Outlook plays a big part in day to day running of the office.

Presumably if they had could switch OSes easily, they would have switched from Windows XP to Windows 7 by now.

XP was released in 2001. If they'd adopted Red Hat they'd have adopted 7, and upgraded 22 times, changing distro once to be on Fedora 20. If they'd adopted Ubuntu they'd have had to wait 3 years for the first release (confusingly numbered 4.10), and now they'd be on release 13. Even Red Hat Enterprise Linux didn't become available until 2002, and you'd be on version 2.1, unsupported since 2009.

If you suck at upgrading and want something that will be supported for 12+ years, Linux is not for you.

Ubuntu's version numbering system is not confusing at all, it just doesn't follow the standard start with 1 and go up. Ubuntu's version numbering system is the date the version got released, so 4.10 got released in October 2004, and version 14.04 (which gets released in 4 days) means that is is released in April 2014. Not so confusing at all.

Edit: Anyway your point is invalid anyway. The reason the IRS suck at upgrading is because they didn't have the budget to pay for said upgrade. Last I heard, it was free to upgrade the majority of Linux distros. Although there may be many other reasons not to switch to Linux, this is not one of them.

There's a lot more to the cost of upgrading than the price of the license. Licenses for Win7 are nothing compared to the price of upgrading all the custom software they developed over the years to insure compatibility with a new OS. The reason almost no business or organism decided to leave XP for Linux is because it's not economically advantageous to do so. When talking about an investment as big as this, you can be sure that almost every options have been evaluated.

Do you live in a world where linux just magically upgrades itself on 110,000 computers for free?

Er ... yes?

http://imgur.com/omXommu

(Except the "110,000 computers" bit, this is just one.)

Who runs this UI on 110,000 computers?

The 110,000 computers bit was the only bit that mattered.

In my federal government research lab, we have 15+ years of Excel scripts behind a huge amount of our ongoing analyses. Whether this route was the best or not is a bit moot now (Excel dependence grew by individual choices on the local level); currently the expense of the switch would be very high. Also, in any interaction with the outside world (like Universities), Office is still the de-facto standard. An attempt of mine to lead a co-authored study in a Latex document, with several scientists at universities, caused mass confusion (as opposed to just 'track changes' in word).

We're slowly creeping to R, python, etc., and Linux is allowed as a choice (about 10% of us choose it). That said, my lab made it over to Win 7 about 2 years ago now...

I like the linux alternative, but I like even more that decisions like this are made at a local level, rather than through a massive top-down disruptive push.

15 years of this Excel? I guarantee your analysis is incorrect. How much is it worth to get correct analysis results?

Eh, cheap dig. We are under specific legal and regulatory burden to produce, proof, and provide our results for external audits, which happen regularly. QC is a substantial part of the costs of a new system. Thanks for the offer of guarantee, though!

There have been billions of dollars poured into government software that runs on Windows. Billions more poured into support, verification and validation of that software.

Switching would be very, very, expensive - and its not clear that it would help even in the long run.

Additionally, Windows is still THE dominant platform for science and engineering work, especially modeling and simulation. How many engineers (non-CS) do you know that do not use Windows? The most popular CAD software, SolidWorks, is available only on Windows. Alibre, the second most popular one, is also Windows only. The next most popular one, AutoCAD, is Windows and Mac only. 3DStudio Max - Windows only.

Most simulation software is also Windows only, at least when I was a government contractor, that was the case.

If the government switches to Linux, they'll have to switch en masse - having some people on Linux and others not won't work very well at that scale of bureaucracy. And when even most engineering software isn't available on Linux, that isn't gonna happen.

I think the most important factor is that Microsoft puts a ton of effort into backwards compatibility, to make sure that it is always easier/less risk to upgrade Windows (and Office) than to switch to something else.

I know that in the corporate world there are still very many business applications that are mission-critical and very special-purpose (sometimes custom) that were not written in a cross-platform way, and may not be maintained anymore - I would be shocked to find out that it was not this way in the government too.

It is certainly not backwards compatibility. Every internal review of Microsoft systems within the government indicates they upgrade whenever possible, even to the detriment of keeping versions that "just work". That is one of the frustrating things about succcesses in deployment of open source: a mindset that still wants to put out the latest version due to features and "coolness" instead of patching and keeping stability that would ultimately save money.

If these are user-facing computer, then the obvious reason is because of what people use and the software they rely on.

Even if they do run Linux, upgrade is still a pain in the ass.

The major problem is that anyone over the age of probably 40 would be crippled by a migration to say Ubuntu. My parents still struggle with basic stuff on Windows even though they use it (Win7) everyday at work. Throwing something foreign at them would account for so many lost hours the initial cost would end up being extremely expensive.

Also Microsoft Office, basically MS's golden egg, is way farther ahead then Libre Office regardless of what people say. All of MS office's services integrate extremely well with each other as much as it pains me to say.

At the end of the day all these people need is Outlook, Word and XL which are leaps and bounds ahead of Libre Office, which doesn't even have a mail client.

Add in all the lobbyist and money going towards the right campaigns and there's your answer.

> The major problem is that anyone over the age of probably 40 would be crippled by a migration to say Ubuntu

I accept that migrating to a new platform would entail an amount of disruption, but I don't see what that has to do with anybody being over 40.

Why should LibreOffice have a mail client, when there's Thunderbird or any number of other excellent mail clients?

I'm not saying it should but I figure it would be easier to fully integrate an application that already used the Libre Office subsystem rather that integrating external mail clients.

One of the biggest reasons isn't just the licensing and individual system, but the supporting systems and processes: certification and accreditation processes for security, ability to only pay for one year at a time due to legalities of budget constraints, and items such as mandated ediscovery systems that are tied to the OS and systems architecture they are designed for using. It takes a great deal of experience, planning, and foresight to decouple services and make removal and swap out of key pieces open and even possible to change out. Add to that active defaults by vendors that encourages lock-in as well as the failure to draw in expertise that would survive in such an environment, equates to mediocrity in IT.

Reply on news.ycombinator.com