Linux in the US Federal Government?
Grech asks:
"I work for a US federal agency
that replaces its workstations and attendant software every 3-5 years.
At the moment, the environment for most workers consists of an
OS, an
office
suite, a UTS60
emulator, and an X
Server. Logic seems to say that when all this gets ripped out
and replaced in a year or so, it could be done cheaper with
Linux, but a case will have to be made, and a strong one. I've got
the arguments, but I need the numbers and the anecdotes to back up
such a huge project."
I believe its a $6B contract, and its in progress right now. I'm not sure if the consensus is that they're being successful or not, but its the biggest single example of a "managed solution" that I've ever seen.
SuperID
The primary impetus for this coming round of upgrades is a new application, the one which requires the X Server to run. The idea behind it is to add a layer of abstraction between the users, and the legacy app from hell, the one which requires a UTS60 emulator, with an eye toward banishing the foul thing. Of course, Exceed is a real pig, as always. Everybody has the office suite, but most only use it to read .ppts and .docs on the Intranet. The rest use it to POST .ppts and .docs to the Intranet.
It may not be just, but it is fair, and that is more important.
and, as much as I'm with you on this one, I don't think you have a prayer.
AFAIK, the only important use of Linux in the IRS is our border routers, admined out of the Indy office. They run RedHat. Get on our intranet and check out the common operating environment specs and list of baseline software. That will tell you what distributions are allowed and will give you a good hint as to where they are being used. Of course, there are lots of Unix servers (two big banks of them within 50 feet of where I sit as I'm writing this), but you asked about the desktop.
There are essentially no Linux desktops in the Service. The Unix desktops in the IRS are rare. Revenue Officers who have not recently had their equipment replaced use a pure SCO OSR 5.0.4 environment, complete with WordPerfect for Unix 5.1 and Lotus for Unix 3 point something. That would be about 5000 users at this point. However, those users are migrating to Windows and there was no budget to rewrite the pile of custom apps they use. Result? Just like the call sites where desktop Unix used to hold sway, these users are going to a Windows-centric desktop with a full copy of Interix for their Unix. In case you didn't know, Interix is the Unix owned by Microsoft. Yes, we buy one of our Unix variants from Microsoft. And, yes, it works about as well as you might expect a Unix to work when it's owned and supported by Microsoft. (Ask me sometime about the Oakland ICPnt/Interix rollout. The persistent connections to the DNS caused by an Interix flaw resulted in the most idiotic work-around I've ever seen: the DNS server for the entire area got a script installed to reboot it hourly. That was the only way to keep it running and they had to do it for days till Microsoft could come up with a re-write.)
Lessee - That means that we're going to have many thousands of users using apps written under SCO and intimately tied to WP5.1 who will be running those apps on Interix which is running on top of WindowsNT 4.01. How does that work? Not very well, I'm afraid.
Roughly the same thing is happening with all the call site and service center employees who formerly had Unix desktops, although the situation in those places isn't as bad as it is out in the field.
Where does that leave us? Where does that leave the prospects for Linux in our organization? The failure of ICSnt and ICPnt (the two main projects involving migrating Unix users to Windows while retaining their old Unix apps) to smoothly migrate users to the brave new all-Microsoft world that our executives want has simply convinced those executives that they were right all along. IOW, they hated Unix on the desktop before, they are having problems running Unix apps on NT now (duh), and thus they conclude that this Unix thing or anything even remotely Unix-y on the desktop is clearly crap. They want it gone. They don't want anything to interfere with their thrice daily ritual of facing Redmond, kneeling, and symbolically kissing Bill Gates butt in prayer to the great god Microsoft.
I can give you just one ray of hope. If Microsoft pushes their licensing schemes forward, we're in about the worst position I can imagine. The money that will have to be thrown away on licenses is *huge*. That kind of budget-buster is the only thing I can think of that would cause the Booze/Allen/Hamilton-Microsoft-worshipping execs at the top of this organization to even look at Linux on the desktop. If you happened to be in the right place at the right time with a demo system ready to go, you might be able to effect the thinking of a critical analyst who might push for a study or two that might result in a pilot project in 3 or 5 years.
Seems like a long shot to me.
ps - If you want more and probably better informed opinions, post this to the Kibbles & Bytes mailing list or drop an email to Scott in Austin. (A note to observers: If you're in IT in the IRS, you probably subscribe to the named list and you almost certainly know or know of Scott. Sorry about the insider references in a post that will be read by a much larger audience.)
The first thing you really need to research is why these systems are being replaced.
If the equipment is still functional and it is being replaced, you might have a valid FWA claim.
Solid state electronic equipment should last far longer than 3 to 5 years. Replacement parts should easily be available.
I know that there is a strong urge to update hardware to run the latest software. But when you look at the tasks performed by most of these office machines, you will find that they are used for basic office functions.
This means that the features provided by newer software largely go unused. As a result the purchase of the latest software was a waste. It then follows that the hardware required to run that software was not required.
Consider also that the new software/hardware often requires (re)training of the operators. Those costs are often wasted as well if the operator has no need for the new features of the new software.
So rather than looking at Linux, GNU, Open Source, or whatever to save you money, look instead at justification for making any change at all. If a change of software/hardware is required, find out why. The requirements for the change will dictate what new hardware/software needs to be obtained. If Linux meets those needs then it can be considered.
I really haven't seen anything new in hardware or software in the last 5 years that would require replacement of an office machine that was in working order. Perhaps some upgrade in local or network storage if that is in short supply.
Otherwise, if you really are interested in saving tax dollars, look into justification and perhaps persue a fraud, waste, abuse investigation.
. Quit playing Monopoly with Bill. Switch to one of many non-Microsoft products today.