Slashdot Mirror
Linux -- Government Acceptance vs. Actual Use
sdriver writes: "As someone who works in the Space/Science/Military field (as a contractor) I thought I'd share some links to how much Linux really is being used. This one is the U.S. Military using Linux. This is another. This one is about realtime Linux stuff for the U.S. Military. Then there are some general arguments about Open source here and here. For those of us who have hardcopies of this magazine will notice most advertisments feature Linux along with VxWorks as a typical target platform. Not the "other" guys."
It's no surprise or secret that for all its tripwire maze of requirements, the U.S. government uses plenty of Free software already. Still, a re-examination of the POSIX requirement would seem in order. Or perhaps the Feds would like to save money by funding the development of an add-on POSIX implementation layer for Linux -- after all, that sounds "good enough for government work." ;)
Posix complaince as opposed to 'gee whizz, it seems to work okay', means that a particular build of software has been tested and certified as compliant. That implies a code freeze. It's an expensive process and involves hiring expensive unbiased testing personell. In the end, you have a frozen version of an OS that is certified as compliant.
It's no longer accurate to say that if Red Hat had spent the money to get their distro certified, that it would be Version 4.2 that would be certified as compliant by now. Enough time has elapsed that they'd probably have Version 5.0 certified.
If a specific version of a specified Distribution is certified as POSIX compliant, that has no bearing on any other version, or any other distribution.
Summation: it's not gonna happen in an Open Source world.
I have to say this study kind of surprised me, because yesterday I finished reading the AFCA report "Assessment of Linux Operating System" put out 20 April. Unfortunately, the study is marked for distribution to US Government Agencies only, or I'd provide a direct link. Anyway, the study, prepared by the Air Force Commmunications Agency Directorate of Technology, basically set out to provide a technical assessment of the Linux operating system -- and provide a comparison to WinNT. They found that Linux is a viable file and print server, applications server, FTP server, Web server, and Telnet server. Also had high praise for Samba. In direct contrast to the linked article above, the people conducting the study noted that "security vulnerabilities are often found and fixed much faster than with other operating systems because of Linux's wide user and developer base". "It is in [Red Hat's and Caldera's] best interest that any suspicious or malicious code found in any application be removed. These companies also inform their users about security bugs..." etc. The report's final recommendations stated that 'Linux has matured to the point of consideration for the JTA. It proves fully interoperable with Windows NT and is POSIX compliant. Linux, however, is not DII COE complient. Linux will have to be submitted for the DII COE Kernal Platform Compliance program." The GCN article seems to say that Linux is not DII COE compliant because it is incapable of being so, while actually Linux has simply not yet been through the compliance process yet. In the meantime, usage of Linux has to be "waivered" - we use it anyway.
Well thank you for ruining all my hopes and dreams.
The other thing to note about the story is that you heard about it only because it had happened before on other occasions.
The headline should have been:
Courtmartialled For Choosing Microsoft!
But instead it barely made the goverment computer news and mainstream media REFUSED to cover it at all. (I know this because I submitted the story as a lead several reporters.)
I seem to know at least as much as you. I know exactly what you're saying, but you made the claim that the OS was not at fault. What facts presented in the article allow you to conclude that? The facts seem to support my conclusion more than yours since yours relies on the "inept reporter" being wrong about the OS crashing, which we don't know for sure. Additionally, it wasn't the reporter who said the OS crashed, he was paraphrasing a memo from Vice Admiral Griffin, apparently using the memo's exact words for that portion. I would assume that the Admiral got that information from someone who was administering the system. But again, none of this is really confirmed. But it certainly doesn't support your claim that the OS was not at fault.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
As others pointed out, a crashing app taking down the OS is the fault of the OS, or at least the OS administrators who let the app run with priveleges that allow it to crash the OS. This again may be the fault of the OS, or at least the designers choice of OS if they can't make the app run without having that level of control.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
If it was just an app crashing, why the hell would they have to tow the damn ship back to harbor to fix it?
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
I don't normally stoop to name calling. But you take the cake. You are a hypocritical idiot. You take the same information, from the same article, and claim in this post that it was not the fault of the OS. If the information in the article is so suspect and cannot be reasonably analyzed due to its likely innacurracy, then how can you make such a claim? But you don't stop there. Now you claim that I am trying to read something into the article and make conclusions based on my personal biases. Give me a break. I made an analysis of the facts as they were presented in the article and from that determined that it was probably both an application and OS failure. For some reason you seem to think that your analysis and biases hold more weight than anyone elses. You should quit complaining about stupidity on Slashdot. From reading your posts, you are a prime contributor.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
The key point is that this was an application failure, it was not an OS failure.
No, the key point is that a crashing application should not take down the OS, and certainly not all the other NT machines on the LAN as well. This indicates an OS failure in addition to the application failure, or at the very least a design flaw of the highest magnitude (which could possibly include the choice of NT as the OS).
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
I'm currently a Department of the Navy systems / network administrator, civil service. We've been using Linux for the past 4-5 years, running DNS services and Sendmail. In the past year, Linux has seen use for Squid caching proxy servers, DHCP servers, SAMBA servers, log hosts, RADIUS servers, and network troubleshooting / analysis / monitoring. In that time, we've experienced only a couple of software-related systems failures, and they were due to misconfiguration that was quickly and easily fixed.
m l for more information.)
As far as server certification goes, our use of Linux is known and acknowledged by the regional IT coordinators. There aren't currently any problems accreditting these Linux servers: they are configured by the same guidelines as is required of any Unix/POSIX system. Obviously there are some differences, but there were already allowances in the process to deal with Solaris, HPUX, AIX, DG/UX or whatever variant of Unix the system was designed with.
While true that Linux isn't currently certified for DoD use, and is operating via a waiver, Windows NT was put into place in a very similar manner over a much longer timeframe, since 1996! Just recently Windows NT received the necessary certification for government use (C2).
SGI is working to bring C2 certification to a Linux distribution by late 2000 / early 2001 ( see http://slashdot.org/articles/00/04/12/1035205.sht
Along with OpenMotif, and the other interoperability enchancements being made to Linux every day, I don't think there will be any insurmountable problems using Linux in the DoD in the future. But that's just my opinion.
-Falcor
Actually Darwin's point was that the most successful organisms end up with the most surviving prodigy. In other words, the bearded Mormon polygamist from the 1850's with 5 wives and 54 children was almost certainly more successful biologically than your ancestors.
Success as Darwin would define it doesn't have anything to do with being smarter, or stronger, it has to do with successfully passing on your genes. You could be of sub-par intelligence, and weak, but if you have lots of children that survive you will still be more "successful" than the Professional Wrestler with a PHd in Astrophysics that decided to stop at one child.
Not that any of this matters, I just don't particularly care if I am off topic.
--
It appears a number of moderators are misusing their privilege to try to prevent the truth from appearing in discussions.
This article about the Navy cruiser was widely discredited as it relates to the problem being a fault with NT.
A divide-by-zero error is an application issue, not the operating system, unless it were to have occured in a device driver or in the kernel itself. Which seems highly unlikely given that the articles talk about the fault being caused by bad input data.
Yesterday I read a post where someone commented on how unfortunately ignorant most slashdot posters are.
This is an example of one such person. The fact that anybody picked up on this story without any sort of critical thinking skills kicking in is highly disappointing.
Even more unfortunate is how it was moderated up, and the posts pointing out how it had already been discredited have been moderated down.
Let me basically summarize the article and the simple understanding one has to use to read it:
Basically the article talks about the Navy deploying a system to automate portions of their ship. The system was built to run atop Windows NT 4.0 server.
Somehow bad data was entered into the database. A zero was entered into some record by the system admin, says the article.
This caused the application to crash with a divide-by-zero error because of a lack of assertions on input parameters in the application.
The key point is that this was an application failure, it was not an OS failure.
What is surprising is the number of people who claim to be intelligent but cannot understand the distinction.
Sigh...
Good grief. Where does it say that?
You are trying to insert something in between the lines.
Nowhere does the article state it took out the LAN... It says it took out LAN consoles and remote terminals.
What does that mean? The article doesn't tell us, but one can imply that these were basically Network computers attached with maybe say a Java app, or a web page, or something.
Come on, you can't possibly be sitting there with a straight face and telling me that you actually think an NT Server going into BSOD would take out a LAN? Give me a break.
This again is a lack of critical thinking skills.
Only soldiers trained for doing the repair work would be allowed there. Nobody but individuals with proper classifications (namely mostlt soldiers again...) would be allowed to work on servers in places like Crystal Palace (SAC/NORAD primary command).
Not all critical DOD servers are in locations where civilians would be allowed to begin with- nor, would you really want them to allow people there. Giving them the resources to do this stuff easier would be a godsend.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
I think we've got a cluster of Linux boxes for compute-heavy tasks somewhere, but we used standalone Linux workstations for all the development on one of my projects this summer, after having migrated from HP-UX (mostly my doing) last summer.
Linux:
Is much, much faster for an affordable box. The HP C360 computers we got a year or two ago were about as fast as PII500 boxes, but at three times the price. We've got nicely parallel code, too, so the dual PIII700 systems they're using now come in handy.
Is easy to port UNIX code to, especially if you standardize on gcc and GNU make to keep the Makefiles identical (or use something like autoconf, I suppose). There were a number of source code changes we had to made, but they were all of the "bug which didn't trigger under HP-UX" variety, not bugs in Linux.
Is much better supported. Yeah, that's right. We've got enough geeky interns like me (and highly qualified Unix sysadmins) around to handle any software issues, and being able to run down to Best Buy for hardware certainly beats most proprietary vendors. HP in particular canceled support for that C360 line, that being the final straw. Of course, this may be an HP problem; we've got Sun servers with an amazing "we'll send a tech over right away with a new board" support contract and no worries there.
Sucks when you run out of memory, in 2.2 at least. I got used to keeping a top window running and a kill command ready when testing fast, leaky code; that's because while I can kill the offending process when the VM runs low, the Linux kernel is more likely to hard kill the X server or do something else that will hose the system to the magic SysRq key point.
Has limitations, but has the source code to work around them. There's a max 4kb SysV message size limit in the kernel, for example, but I modified an IBM employee's 2.1 kernel patch to make that limit runtime configurable.
Comes with desktops that are a lot more enjoyable to work on. Some of us booted to KDE, some to Gnome, but everyone was quite happy never to see a CDE panel again.
Scales much better than HP-UX. We're working on prototype satellite groundstations, and being able to put everything on a ThinkPad blows away the anthro cart "portability" there was before. Hell, how much RAM does an iPaq have? It has enough CPU power to run our software, anyway.
I agree that it makes sense to promote free unix for government institutions, I'm not sure that Linux is the right thing to push. Normally, I encourage people to use Linux (over both non-free and BSD systems) for a five main reasons:
1. Linux is free in cost
2. Linux's source is available if that's your thing
3. Linux is politically subversive
4. Linux's community is folksy and congenial
5. Linux is more bleeding edge than other OSes, and is thus more fun
For PEOPLE, these makes sense. For schools, these makes sense. For the government, only points 1 and 2 makes sense. The government wants a free unix that has available source, but they don't want to have to release that source to users, they want a support comminity that is fastidious rather than hip and they want stability and unexplotivity over all else.
What would I reccomend for a lot of government institutions? BSD. OpenBSD, preferably, since it's got another government-friendly plus: security.
Don't get me wrong. I'm as big of a GPL pusher as the next guy, but I've come to the conclusion that the government is a lost cause as far as Free Software goes. No matter how hard we try, CIA agents won't start attending LUG meetings (except perhaps to keep an eye on the Red(hat) Menace), DoD personnel won't post patches to Linux-kernel, and NSA spooks aren't going to release Enlightenment themes any time soon.
No, I think as taxpayers, the best we can hope for is that the government will stop adopting $1,000,000 solutions where $0 ones will do nicely. And as much as I'd like to wish that Linux is the best tool for the job, a lot of the time it isn't. What makes me like Linux so much - that it's got all kinds of quirky features and kernel development is so fast, also does make it less stable to some degree. As a user, I not only accept that trade-off, it's what gets me using Linux in the first place (a fun unix, whoda thunk it), but for the feds, they want something more boring. And slow-but-steady OpenBSD seems just about perfect.
So what's the problem? Not enough POSIX testing? Done on the wrong phase of the moon? Completed but nobody would sponsor the actual certificate? If I was a Linux co in a position to sell to the military, and "proper" POSIX certification cost (say) $US20,000, I'd being sending the money today.
Got time? Spend some of it coding or testing
In every discussion of this infamous event, it seems that one important detail gets overlooked:
The failure of a single computer disabled an entire warship.
Let's be honest: if you can disable a warship by disabling a single computer, it doesn't matter what OS the machine is running. DOS, NT, Macintoshes, Linux, *BSD, Solaris & OS/390 are all equally vulnerable to a lucky shot that causes physical damage to the hardware.
The solution? Either multiple computers, or a distributed OS that can continue to run processes even while individual units are failing.
'Nuff said.
Geoff
I think I see a trend here. Maybe for them it really would be easier to muzzle the entire internet than to produce p
When you look at systems other than PCs used for standard office applications, there is a wide variety of hardware and operating systems in use, including open source operating systems.
Mea navis aericumbens anguillis abundat
Wrong. There are large numbers of PCs running NT at NASA, although Windows 9X is probably more common on desktops. The standard PC desktop software package is Windows 9X/NT, Microsoft Office, Netscape and Eudora.
Mea navis aericumbens anguillis abundat
It's not like the GPL requires them to distribute their modified source code outside the DoD -- only to people using it within the DoD, who presumably already have access to the source. And besides, there are a million ways to frame a national-security exception to complying with the GPL if it did.
The GPL requires that you make the sources available to those who use your version.
If you make some chages and distribute that modified version to some people, you must also make the patch/full source available to them... but nothing requires to make this available to other people.
So, DoD may make changes, distribute it to Navy/... and only distribute the patchs to these, thus, not releasing the modified version/source to the public.
When I was in the Air Force, the very first webserver that was sent to us from headquarters was a Linux box!
This was back around 1993. I remember it well because I set it up at our base.
I guess as long as everyone stays quiet about it, then it is ok, right? *nudge* *nudge*
-- Windows security? Sure, which ONE would you like? -me
Part of the problem is the Government (military in particular) loves to specify (dictate) odd standards and requirements. This is why you can go to a military surplus sale and see barrels of 49 pin swizzle-twist 7+1 notch connectors that were speced for exactly one project (probably the space toilet). The only reason industry puts up with it is because they order alot *and* pay alot to make it worth their time.
This is also true of software. Why use COTS when you can spend another couple hundred K and get 2 extra features?
Maybe if Red Hat starts a GSA cu5t0m m0d division, that will satisfy their needs.
No. It means that at least you have the option.
Ever run an NT shop? You don't just apply service packs or upgrade to the newest version 'just because it came out'. You do it when it is necessary to maintain your level of service, or to fix security reason.
And you'll get a whole lot more detail out of linux than you will any other OS.
We're talking large servers here.. why on earth would you 'want to upgrade because a new version is out'.
There was a situation like this with some guy modifying Doom GPL code (or something like that, I forget the exact situation).
Sigh. Of course it's an advantage.
First of all, at least with Free software it is possible to audit the source, whereas with proprietary software, it probably isn't. That's advantage #1.
Now, of course most users don't have the time or technical knowlege to audit source code themselves. So? The point is, as long as some people do, and if, as a responsible system administrator, you keep track of what they find, you get the benefit of their work. With Free software, many, many people have the source and some of them will audit it.
On the other hand, makers of proprietary software don't usually search for and correct security problems until they have already been found, and often exploited. Very few people have the source so it is unlikely to be audited. That's advantage #2.
Even better, a bunch of companies and/or government departments could form a consortium to fund auditing, and all share the results. With proprietary code, that would be extrely difficult. For example, I'm sure that the US Navy has access to the Windows NT source code, but so what? As you point out, they may not have the time or expertise to really audit it. And even worse, they can't easily form a community group to support auditing, because everyone would have to have access to the source, and Microsoft just isn't going to let that happen. Advantage #3 for Free software.
Finally... you claim that Linux is riddled with major security problems. Well - that's a questionable assertion. Furthermore, it is common knowlege that NT is worse, despite the fact that Linux ships with far, far more capabilities than NT does. Don't believe me - just check BugTraq. But don't make the mistake that Fred Moody did a couple of weeks ago, and add up the numbers wrong. That would be really embarassing.
Torrey Hoffman (Azog)
Torrey Hoffman (Azog)
"HTML needs a rant tag" - Alan Cox
Windows NT got through the back door via a "Posix subsystem" on the kernel.
And, interestingly enough, one of the things on the Air Force's checklist to secure an NT server is to disable the POSIX subsystem. It is apparently a security hole pretending to be a kernel component.
So, the DoD mandates that you have POSIX, but also mandates that you not have POSIX.
Everybody got that?
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Having just quit my job as a defense contractor, the largest problem is getting the local DAA (Designated approval authority, or something like that...) to approve the use of the software.
Pretty much. Taking a look at the big picture, it pretty much comes down to one thing: The DoD loves paperwork. They thrive on it. Whether or not the product you're choosing actually works or not doesn't matter, as long as it has all its certification paperwork. This isn't likely to change any time soon.
This tends to conflict with the open, rapid, community-effort, bazaar style of development. The rapid revs and loose organization don't lend themselves to certification. Even if they did, most certs require you to retest for every change. Since many of Open Source's strengths stem from the rapid revs, this makes things difficult.
And, yes, I realize that you could "freeze" a particular distribution, certify it, and then rev less often. But again: Taking full advantage of Open Source means you have to accept rapid revs.
This isn't to say things are impossible, or that we should give up, or anything else. I'm just pointing out a source of conflict.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
NT really DID make a navy cruiser dead in the water for 3 hours. Some will say it was human error. But the "human error" consisted of entered "0" into a data field. That should be handled gracefully. Some will say it was an application error--the app should have handled the "0". Again, true, but a crashing app should not take down the operating system. NT was at fault and there is no getting around it.
--
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
[1] The reason that the Navy used NT was that federal law states that they must go with the lowest bidder. They had specified POSIX-compliant, because they wanted a Unix system. Well, NT is technically POSIX-compliant, so some !^#*$ company put in the low bid with NT as the underlying OS. If this ever causes any problem in battle, I'd like to see every single one of them shot. What sort of inept moron would put NT on a piece of vital equipment?
Hell, we don't even allow NT in our DMZ...
[1]Surface Warfare Officer, i.e. the most numerous branch of Naval officers
I've seen Linux used in the Army to do things that were just not possible with M$ products, cost being a BIG factor. With all the cutbacks with the military budget, it's just not possible for units to spend several thousand dollars on software. Or the several thousand dollars on licenses. When people (i.e. officers) are shown how flexible linux is (not to mention free), it's not uncommon to hear them make suggestions to problems by saying, "Well, can't you just download something off of the internet?"
Go not unto/. for advice, for you will be told both yea and nay (but have nothing to do with the question)
One of the DoD's biggest problems right now is that they cannot use truely complex systems for highly classified work and still be very certain that there are no security bugs. NT is obviously a lost cause in this respect, even with hordes of people at the NSA with nothing better to do than read NT source code and write buffer-overflow guards. On the other hand, Linux and the BSDs offer a new option: let millions of people pour over the source to find problems, and then pick and choose the most solid and secure programs.
Obviously, you would still have your screening process which would involve an intense security code-review. Not really a problem.
I think in the end, you would still have a linux.mil distribution, but it could track current development MUCH faster than with any other OS. The real trick would be convincing the pencil-pushers that contributing their security fixes back to the projects would be a good thing. Everyone who has had to maintain such a thing knows that you don't want to be stuck maintaining even a 5-line patch over more than 1 or 2 versions of a large piece of software, but most of the people setting policy have never had to do that.
The thing that's going to kill Open source software usage is, though, is the people who think that any program written under Linux must be GPLed. Even *BSD ships a lot of GPLed software, and thus hackles get raised. Someone needs to take on the role of championing GPL PR. I don't mean this is the sense that the GPL should be pushed over the *PL of the week, but that the misunderstandings about what the GPL means should be countered.
-
POSIX compliant means it's been certified as such. That means that people have to audit the code and suchlike.
You get your code audited, you get a pretty certificate to hang on the wall, your code is certified POSIX compliant. The code does not change. It's not like, before the audit it wasn't compliant to the POSIX specification, but after the audit it is. The code does not change. Either yout code is compliant to a set specification or it isn't. All the auditing in the world won't change this.I'm sorry if I sound a little flamey, but what the hell.
G
- Still, Torvalds had high ambitions. He was writing a toy, but he wanted it to have many, if not all, of the features found in full-strength UNIX versions on the market. On July 3, he started wondering how to acomplish this and placed a posting on the USENET newsgroup comp.os.minix, writing:
- Hello netlanders,
Linux has been POSIX compliant since it was a hack project in Minix. But compliant and Has-Paid-Us-Lots-Of-Fees-And-Is-Certified compliant are two very different things.Due to a project I'm working on (in minix), I'm interested in the posix standard definition. Could somebody please point me to a (preferably) machine-readable format of the latest posix rules? Ftp-sites would be nice.
G
It doesn't really make a lot of sense in this thread.
Sorry.
G
-
Wrong. Linux has aimed to be POSIX compliant. It's close enough that most people won't notice, but until someone has poured over it, you can't be sure.
You boldly state that I am wrong, but you do not go on to back this statement up.Compliance to a specification is like a scientific hypothesis: you cannot prove it true, you can only prove it false, by finding an example where it fails.
Going through the rigourous POSIX compliance tests is like testing a scientific hypothesis, and you can understand why many scientists would choose to not accept a theory as being valid until is has been scientificly tested.
But the mere fact that a scientific hypothesis has not yet been tested does not mean that the hypothesis is not in fact correct. And just because Linux has not been tested to ensure that it implements the POSIX spec. correctly, does not mean that it does not actually do so.
(POSIX compliant) != (certified POSIX compliant)
It is possible that there is a fault in Sun's implementation of POSIX in Solaris. Testing cannot possibly hope to check every possible combination of variables. It is possible that the Linux implementation is absolutely perfect. Having Linux tested would be nice, but is pretty impractical for an OS growing and changing so quickly.
Linux has not been put through the standard set of POSIX compliance tests, but the set of functions that are required by POSIX are put to test by millions of machines every day. I have not heard of there being any known faults in Linux's POSIX compliance. If you know of any faults in Linux state them. Otherwise, you do not flatly state that Linux is not POSIX compliant, as you do not have any evidence to back this statement up.
G
-
The plural of "anecdotal evidence" is not "data."
True, but it is now you who is doing the pedant-stylie hoop jumping.What does an OS have to do to be compliant with the POSIX spec.? It has to support a variety of functions, e.g. fork, and implement them correctly. How many times do you think that the fork instruction has successfully executed on Linux boxes around the world, in the time you have been reading this post? Are you actually suggesting that the fork instruction is not correctly implemented, or are you just playing word games?
There are two ways that you can really test code. Check it in theory, by independant code review, and check it in practice, by running test data through it. I would suggest that:
-
The core of the Linux kernel is constantly under a far higher level of scrutiny than I'm sure would be necesary to complete the POSIX certification process. Many eyes make bugs shallow. And opensource software has a lot of advocates in academia - very intellegent and experienced computer scientists.
-
No testing process can test every every possible combination of variables. Linux is being put through its paces live on a hell of a lot of machines around the world, and is passing the test. Data? No. But a lot of anecdotal evidence.
Anyway, although it carries less wieght, "anecdotal evidence" is a form of data. I [personally] have no firm evidence at all that there is such a thing as gravity. I haven't performed any tests. I only have anecdotal evidence that I cannot fly. But I have a hell of a lot of evidence - experience from every day of my life. This is not proof, but then nothing can ever be proven, only disproven. It is firm enough for me to choose not to jump off tall buildings.I believe in gravity until someone demonstrates otherwise, and I believe that Linux is POSIX compliant until someone demonstrates otherwise.
Smile,
G
There is no security compromise here.
Anyone can hack Linux but they can't force you to run their version. The official kernel and distros are the ones you'd stick to and are probably safer in terms of scrutiny, just look at the recent MS "netscape programmers are weanies" debacle. A security hole (back door!) right there in the proprietary closed source system.
As usual the security issue is FUD. Any security expert knows that security through obscurity is no security.
But since the GPL only applies if someone redistribues or sells the software, this probably wouldn't ever happen. I mean, why on earth would the DoD redistribute the software with their changes in it. According to them, that would probably be a threat to national security.
Though, the government has been known to do shadier things than sell software and break a public licesnse to raise a buck or two...*shudder*
First: This article is dated 1998. Now I know that technology has not advanced that quickly in the past two years but I think that that discounts the idea of the system being a Java app, or a web page .
Second: The Battleship's hardware system (like the missile launch system for instance) is very likely not going to be a part of the NT operating system hardware list. Ie:
Device Manager:
- CDROM
- Disk drives
- Display adapter
- Hard disk controller
- Infra-red missile tracking
- Keyboard
- Missile launch/guidence
- Monitor
- etc
Therefore the Battle systems are most probably controled through a proprietry software system that runs on top of NT and uses NT for communicating. Bringing me toThird: I find the idea that a Battleship's main system being run by a large number of independant and separate programs is highly unlikely. All the systems would need to be synchronized and interdependant to operate in a realtime enviroment. The missile systems need to get targeting data from the radar/tracking systems etc. This would all work in a combined effort. I would envisage that the whole system would be operated by a central, intergrated master system and the individual controlers (ie human users) would be running a remote client that interacts with the realtime system. A bad bit of data in the master system (an application) could result in the entire system being crippled. At some point in the system, everything needs to be synchroized and that also means that that point is a potential show stopper for the whole system.
This is not a case of someone trying to view dynamic VBScripted content in netscape, and crashing the ship. This is a case of the system (Ships system, not the individual OS) being controlled by a grand application (the battle system) and that application being disabled by bad data. Kind of like when your perl script CGI trys to put text in a number field of your backend database and takes out your entire dynamic website. The operating system is ticking along fine but the system (the web site) is dead in the water.
That said I still think NT is a pain in the a$$. But I would like to condemn it for the the things it is guilty of, not just what I can pin on it based on olympic conclusion jumping. (Hmmm. Demonstraion sport at Sydney perhaps? Where's SOCOG's number.)
"I'll take the red pill. No! Blue! AAAaaaahhhhhhhhh"
- Monty Python meets the Matrix
HERE'S AN IDEA!
Why don't you READ what he is SAYING and apply some critical thinking.. there was one server.. the other "terminals" on the LAN were probably all using that app. The app crashed, the ship went dead, because everything was using that app. The "LAN" didn't really die, but the app that everyone used did. This is the conclusion any rational person (who knows something about computers) would come to after applying non-ideologically colored logic to an article that is -- again to any said rational, computer techie person -- written by a technologically clueless journalist talking to average end users about the incident.
GEEEZ! I know this is flamebait, but I got so mad reading this damn thread..... I can write a shitty app in linux, tie a ship together with a LAN, have the app crash (NOT take down the OS), and you know what? none of those terminals will work there either. Nothing on the ship (computer wise) will work. Does that make linux a shitty OS?
---
DO NOT DISTURB THE SE
This raises an interesting question about the GPL: if person A modifies GPL code and gives the binary to person B only, can person C complain about it? If not, then what's to stop a person taking GPL code, modifying it, and giving the binaries only to people who are willing to sign an agreement where they promise never to ask for the source code? That would effectively let person A distribute binaries without having to distribute the source code.
Imagine buying a piece of hardware with an embedded version of Linux. When you open the box, you inherently agree to a license. Part of that license says you agree not to ask for the source code.
--
And the men who hold high places must be the ones who start
To mold a new reality... closer to the heart
http://www.goingware.com/decss
We now return you to your program.
-- Could you use my software consulting serv
Isn't Linux already POSIX-compliant (or at least mostly, anyway)? Why would an add-on POSIX layer even be needed?
Disclaimer: The opinions expressed are not necessarily my own, as I've not yet had my medication today.
It's a good thing that they are using more free software, because anyone who's ever had work on DOD stuff knows that a) Copy Controlls are Frowned on and b) your tech support records will show far more users then you actually licensed them for.
Which is why companies like IBM and HP special GSA notices on all of there software.
You may laugh about how kludgey those ancient tape drives are, or maybe about the 10MB hard drives the size of washing machines, and yes, they're old and slow. But when the ship takes a missile or torpedo hit, or an electromagnetic pulse, these slow and old systems keep on running. You won't see a BSOD as the clunky old hardware transmits the launch signal of death to the missile launchers.
Go Navy! Beat Army!
Of course, we're now using Linux for the usual SAMBA and webserver stuff as well.
This link to the Avalon system at Los Alamos may be reachable from the outside. I'm inside the firewall right now, so I can't verify if the link will work for the rest of you (it works for me). There is another relatively new Beowulf cluster at Los Alamos, 128 dual processor P-III's, IIRC, named Rockhopper. Sorry, don't have any links for that machine. Some of the folks who put that machine together now work for the TurboLabs division of TurboLinux.
The above article is excellent.
It can be summed up like this: "If we used Unix we wouldnt have this problem."
hehehehhe
peas,
-Kabloona
I would imagine in places where security is an issue, the government should be looking at BSD first. Not to diss Linux, but OpenBSD is reknowned for it's "security by default" out of the box. If anything, I would think the government would err on the side of security (so their government hire doesn't get the bleeding edge driver or graphics utility, boohoo).
Can you just imagines the news?
Reuters: This just in - The US government has decide to standardize it's military computers on an open source operating system developed mostly in Canada. Interestingly enough, "OpenBSD" as they like to call it is based on the work of some computer scientists from the University of California at Berkeley a couple decades ago. Apparently everything the US has to offer pales in comparision when it comes to the security of OpenBSD. An admiral that wished to remains anonymous said, "It's about time those canookies and hippies did something to watch our ass. After all, we've been watching theirs for all these years."
My question is does giving it to other organizations, (Army, Navy, etc) would be distribution.
One can make a good argument that the answer is 'yes'. When I worked for a U.S. Navy laboratory, we couldn't, for example, get a license for unlimited use of a software package for the lab and then let the entire U.S. Navy use it. There's some point at which you're dealing with a separate entity, even if it's within the overall organization. Perhaps someone familiar with the GPL could elaborate on how it defines the boundaries.
Windows NT got through the back door via a "Posix subsystem" on the kernel. Using this subsystem you can boot into the Posix subsystem but not use the other Windows NT (2000) features. The Posix subsystem uses programs from such other vendors as Mortice Kern Systems to give it Posix compatibility. DEC's VMS used a similar technique.
But just because an OS has a Posix subsystem does not mean that the applications that run on the other subsystems are Posix-compatible at all. Microsoft Office is not, for sure.
In addition, many computer systems used by DOD are supposed to have a certified level of security--C2, even B1 in some cases. Windows NT did get C2 certification, but only for the NT 4.0 version, not for the 5.0 version many departments bought. This includes both hardware and software in one security level.
No doubt RedHat or VA Linux or IBM or Compaq could cobble together a GNU/Linux or OpenBSD or whatever system and pay the money to have an independent body certify it at the appropriate security level. At least one Posix-certified version of Linux has been released. Then DOD could buy the Linux or BSD OS and hardware from an approved list.
I worked for the Navy as a contractor Unix sys admin. We had on our base about a thousand SCO Unix clients running on Everex-type PCs. The idea of the managers was to dump SCO and go to Microsoft Windows NT. Linux could have been chosen as the client instead of Windows, but it would have to be locked down.
The huge Solaris enterprise systems that ran the base used Oracle databases and ran many old Cobol programs too. It's possible that Windows could run those programs, but they would need a great deal of rewriting. Linux as yet probably doesn't have the beef to run the big Navy RAID systems and databases--but Linux will get there soon. Maybe IBM will port CICS to Linux--it's already on AIX.
The reason for "standardizing" on Windows NT was that it would be "uniform" across the network. A stupid reason, and one that will cost the Navy lots of money. For example, there is no reason to give many of the users a whole Office platform--all they do is enter numbers into a database screen in their job. On the other hand, getting Windows email working right will be a nightmare for such a huge system. The big brass don't listen to the users any more than Billg does.
There are a lot of Linux users in the military, and they would love to talk some sense into the brass on this subject. Government Computer News has been educating them for years. But military brass are the last to regain any common sense.
It means you have to take responsibility YOURSELF for ensuring that you stay consistant with whatever setup you want.
Is that supposed to be an advantage? What you're saying is that I'm supposed to audit millions of lines of code if I want upgrade to a newer version of RedHat or pick-your-distro.
I don't think it's reasonable to blame the users when a Linux package is released with major problems, particularly major security problems (which, unfortunately, Linux is riddled with).
--
Sometimes it's best to just let stupid people be stupid.
Seriously, I'd think that national defense might trump copyright?
With proprietary software they could be in the same situation again if the vendor moved to some unfriendly country.
A valid point, underscored by Toshiba Machine's, gaff several years ago, of actually selling silent submarine technology to Russia. People have been executed in the US for less.
Vote Naked 2000
A feeling of having made the same mistake before: Deja Foobar
No, no, you got it right, the trick is the Rosenbergs should have:
Contributed in a big way to Ike's campaign
Been incorporated (It wasn't us it was the corporation! Kill it!)
Vote Naked 2000
A feeling of having made the same mistake before: Deja Foobar
The federal government can't afford to pay compeditive rates to IT professionals. So, they pay lower rates to second-rate employees. These employees aren't smart enough to learn Linux yet. Perhaps in a couple years when Linux is easier to use, the DOD will adopt it.
I personally tried to get my section to convert our web server to Linux, but we're sticking with NT because that's what our people know.
In case you're wondering, I'm planning on quitting once I graduate. Until then, it's good experience and looks good on a resume.
...considering the pool of computer people the government employs. After having worked at a DOD contractor for a few years during college, I realized that many of the people who work there have no clue about anything technical involving a computer. Too many managers, not enough techies.
...because the government shouldn't have shackled themselves to such a proprietary system as Windows in the first place.
I mean, I understand that there weren't many open systems back when the DoD, for example, started migrating from MS-DOS to Windows for servers. However, there have always been operating systems where you could at least view and audit the source, if not give it away.
IIRC, AT&T Unix was always fairly "open" in the sense that you could contract with them to look at the code. Somehow I doubt that the government ever got that option from Redmond.
Dewey, what part of this looks like authorities should be involved?
From the ma-linux tux list:
l
BTW, there's actually another reason: All gov't computers are required to
be fully POSIX compliant, which MS OS's are not.
"...shall have a POSIX compliant, multi-user, multi-tasking OS [operating
system] that is capable of providing the following services concurrently:
print, file, communications, networking, and database."
I've brought this up lightly at the Dept. of Ed. and people didn't care.
-brian
On Thu, 27 Jul 2000, Przemek Klosowski wrote:
> Government Computer News column by S.M.Ryan:
>
> http://www.gcn.com/vol19_no20a/opinion/2449-1.htm
>
> brings up the topic of a Federal Acquisition Regulation Council rule,
> known as the 'blacklisting rule', that requires the government to
> do business only with vendors that it finds 'responsible'. He writes:
>
> The proposed standard remains the same: no finding of
> responsibility for vendors with "an unsatisfactory record
> of integrity and business ethics." But the new proposed
> rule requires contracting officers to rely more on
> objective measures, such as findings or decisions by U.S.
> courts, administrative law judges and boards of contract
> appeals.
>
> Could some enterprising contracting officer therefore
> conclude that Microsoft, the world's largest software
> company and a major seller to the federal government, has
> an unsatisfactory record of integrity and business ethics
> based on Judge Thomas Penfield Jackson's detailed ruling?
>
> I think the answer is yes.
>
> If this comes to pass, it would be quite explosive, and would
> certainly be a good argument for the Supreme Court to decide the
> Microsoft case quickly.
>
>
> p
>
>
> NB for the legal eagles among us, I think the crucial passage is 52.209-5 (D),
> page 40833 of Federal Register Vol. 65, No. 127, Friday, June 30, 2000
> and can be obtained via http://www.arnet.gov/far/ProposedRules/99010.pdf
>
--
Brian C Merrell
merrell@tux.org
BilldaCat
People have to get over the 'anyone can modify it' stuff.
'Anyone can modify it' means two things.
1) The source is available
2) It's LEGAL to modify it.
Nothing more. It doesn't mean that 'anyone can walk into your installation and 'modify' your stuff.
It means you have to take responsibility YOURSELF for ensuring that you stay consistant with whatever setup you want.
sheldon wrote:
Do you include yourself in this list?
While it is true that a typo killed the application, the application killed, not just the computer it was running on, but all of the computers needed to run the entire ship.
Now, is that an application error or a system error? Who knows? I wouldn't think that an error in a single application would be able to take out an entire LAN, or even the computer on which it was running without some help from the underlying operating system.
That is why I am dissatisfied with the explanation that the error was an application error and the implication that it would have happened under any operating system.
I work as a Sys Admin for the US Air Force (SrA.) and I use Linux as a print server and internet connection firewall. There is even virus software (McAfee) for download at the site where I download updates for my Win NT/9x machines. Linux has not been "officially" adopted yet but like many other places, it is in use. The military does not have an "on paper" official OS that we must use but we are encouraged to use NT because it is more "secure" (wonder who made that assumption, I bet not a sys admin) but we can use Linux if we want.
"If ignorance is bliss, why aren't there more happy people in the world?"
WWJD -- What Would Jimi Do?
I am quite civilized, and I should be brought a beer immediately. -- Bruce Sterling
It givesa a whole new meaning to core dump.
The warp core, she's dumping all over me cap'n, she can't take much more, she's gonna blow!
Linus reserves the right to break Posix when he thinks it's being brain dead. Also, bits of POSIX are still in the works (Like the Linux Privs stuff.)
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
Although the Yorktown did not have backup systems, Redman said that future Smart Ships will have systems redundancy to ensure that ships can continue to operate.
I can see it now.
[Primary system failure]
BSOD- Divide by zero error
[Switch to backup NT systems]
BSOD- Divide by zero error
[SDIW- Ship Dead In Water]
Commander: "What the... Fucking Windows NT! Get Bill Gates in here."
[Bill Gates enters]
Commander: "You told us SP6 would be more stable, faster and have better failure protection."
Bill: "It does, it's over 400 times less likely to.."[BLAM, Commander blows his head off]
Yes I shamelessly stole this scenario from the Southpark movie.
Steven
-- I have marked myself unwilling to moderate-- I don't have other accounts to artificially inflate the karma of
Security can't be their primary concern, heck, the DOD has been cracked so many times it's a wonder their not on a firs name basis with their guests.
/usr/home4/pixarc2/"
"Hi, Bob, How's it going?"
"Oh, Pretty good, yourself?"
"Can't complain, got a new DSL, just trying it out."
"Nice, what are you looking for?"
"Oh, just the usual, some prOn pix of the Joint Chiefs."
"Look on
"Thanks! Catch ya later, dude!"
"No prob, say 'Hi' to 31337 7rO11 8OY for me."
Vote Naked 2000
A feeling of having made the same mistake before: Deja Foobar
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
I live in Europe. Here, there are plans to ban MS Windows from government key networks. The two reasons behind these projects are the following (and they do apply in many other situations) :
- avoid backdoors intentionnally placed by the software manufacturer (code can be checked for backdoors). Think of a world where Microsoft keep the US Government in hostage "you stop pissing us off with those anti-trust trials or whe shutdown all your NT system...". And, with the UCITA, they could even have the right to do it !!! Not speaking of information leaks (spying) by the various backdoors (or data corruption ?) Risks from a closed-source system are far to high or governments !
- vendor independance. Think of the case here Microsoft stops to make MS Windows products. All updates stopped, no more bugfixes, no new licences for the new machines to be incorporated in the network, no new-hardware support,... If that ever happens, the system would have to be changed fast, involving many costs (reinstallation, porting of in-house programs to the new environment, buying of replacements for some programs,...) Such a transition would be very disruptive if it has to happen from one day to the day after.
Beside this, the availability of sources is very interresting for military purpose. It allows them to put in their own modifications to harden the security, to cut off unneeded parts (to avoid the security problems into these parts), to hardcode some usually dynamically fixed values (IP address of the NIC for example),... many advantages not provided by MS Windows.
And here is the correct link to the article...
--
*Condense fact from the vapor of nuance*
25: ten.knilrevlis@wkcuhc
*Condense fact from the vapor of nuance*
These are the guys who, aside from inexplicable decisions like considering the F-16 a suitable replacement for an A-10, put Windows NT in charge of a cruiser.
The ones in charge clearly don't have any clue whatsoever.
Besides, do you want free software considered to be vital military equipment? That's kind of a scary thought. A little close to the "munitions" argument over encryption software.
---
Despite rumors to the contrary, I am not a turnip.
Don't let the hoopla fool you. The US Army currently uses Linux as the OS for their Warlord Notebook Intelligence Analysis System.
It is a light weight, cost effective counterpart developed by Mystech Software (later Sterling Software and now I have no clue) to compete against ILEX's All Source Analysis System - Remote Work Station (ASAS-RWS or just RWS for short). The RWS sits on Sun architecture and Solaris 7. The RWS code from ILEX really sucks too.
So, Mystech/Sterling developed the Warlord Notebook using Government Furnished Information (GFI) derived software and MySQL on top of Red Hat 4.2 initially before migrating to Red Hat 5.0 and then 6.0. It was cheaper (by far... as in free! The Army only had to pay for contract support and that was even optional) than the RWS system which was proprietary code from ILEX (which sucked), Oracle 7i (which rocked but at $2k per licence) on top of Solaris and were forced to accept support contracts. The system proved to work so well that some units dropped their RWS's completely for Warlord notebooks. There are a few other factors involved in that as well (such as a laptop notebook vs a Sparc pizza box and monitor... what a weight difference!) and it was generally liked by the users (unlike RWS).
But... money and politics play out and at last word, ILEX was gaining the upper hand again and Warlord Notebook was going to phased out. Not due to performance or cost, but due to the fact the Dept of the Army had spent SOOOO much money with ILEX that they didn't want to throw it away.
It's a crying shame.
I don't drink because I have to, I drink to stop the voices in my head!
I don't drink because I have to, I drink to stop the voices in my head!
Having just quit my job as a defense contractor, the largest problem is getting the local DAA (Designated approval authority, or something like that...) to approve the use of the software. The DAA has the last word on what is and what is not allowed at a site. On some bases, each building has its own DAA, and thus the policy can vary from building to building depending on who the DAA is.
:)
:)
I had to jump through more hoops just to get Perl installed on a few systems that we were using for development work. These were for systems that were not connected to anything outside of the room that they were in (they were classified, but only at the 'secret' level.
The argument being that all 'freeware' is not permitted by USAF regulations. The problem was that the regulations were talking about binary only distributions, not source code. I had fun pointing out that the main AF Publications web server (the one that holds all of the regulations) was running on apache and linux
I eventually got around the problem by purchasing the O'Reilly Perl resource kit, so it was now 'purchased COTS (commercial, off the shelf)' software, and that is ok. I could sit at home, burn this stuff to CD's and sell it to the government for a ton of money... That was ok, but I could not download it for nothing. Your tax dollars at work.
Like I said, this will vary widely from location to location depending on the local DAA and what they are willing to accept the risk for. Unfortunately, most DAA's are GS-14 or 15 (top of the civil service management ladder) and lifetime civil service employees, and thus do not have much real world experience or knowledge.
But anyway, I'm much happier now as a consultant for internet startups where OpenBSD and OpenSSH are accepted. Now I can do my job and really secure the systems
I've worked for the last year on a DARPA program for wearable situational-awareness computers for the military (the Army, specifically). The system we're designing, while still proof-of-concept, is built around Linux, which was chosen primarily for its openness and flexibility.
Parts of this system, probably significant ones, will make their way into fielded systems in the next ten years. Also, there are a great number of DARPA research programs that involve Linux. In other words, the *future* military systems are being developed *now* under Linux.
It isn't big yet, but I'm willing to wager that because of today's research, the next generation of military software will be Linux based...
Let's try not to let fact interfere with our speculation here, OK?
Is anybody else tired of hearing this? "Linux isn't right for us, because of all the hackers out there that will crack our machines". I was with them through POSIX, Motif (God knows why), NFS and CDE, but they lost me at posting the source online. As if /usr/src was openly accessible over the network by default.
Isn't part of the Open Source Creed the Right to Fork?
Mebbe someone with better GPL knowledge can say whether or not the DOD would have to release the source if they made modifications to the kernel to make themselves happy.
Or, I suppose, they could go with FreeBSD, and use the BSD license.
Potato chips are a by-yourself food.
Lots of places in the DOE use linux. A lot of that is scientific computing. Look at Avalong and Loki at Los Alamos. Look at Los Lobos at Sandia. Look at all the clusters all over. It's cheap cycles.
I work as a sysadmin at a DOE facility. We recently decided to include Linux on our unix supported system list, set up a program to buy linux boxes, and started getting a lot of interest. For the stuff we do, a $5k linux box will equal the $30-50k suns and sgis. Before you freak and say I am wrong, it works for us. For our applications. Most of our stuff does not have a 64 bit requirement, so that's not an issue. Basically, we just found what we needed to integrate into our environment (shared filesystems, main applications, etc), made sure our config stuff was cross-platform (PERL is the language of the gods), and put it up. We now have guys who can do work on a $10k machine (including a huge monitor) that would normally have required a $80k machine before.
Linux is breaking the paradigm. Scientists jsut like to see the numbers. You say, "Benchmark this box vs your workstations". They are shocked with the results. For the longest time, sgi and sun (and to a lesser extent hp, but I still love PA-RISC) dominated the market for a reason. Then, more recently they dominated the market because of reputation and past.
Face it, it's hard to beat an x86 cluster for computationally bound tasks for the $$. As a desktop, a $5k x86 box is _insanely_ fast. _INSANELY_. Ask any of my scientists that say "Hey, can I borrow your box to run on? It's just faster than this *insert traditional unix box here*".
-- Who is the bigger fool? The fool or the fool who follows him? --
I would imagine in places where security is an issue, the government should be looking at BSD first. Not to diss Linux, but OpenBSD is reknowned for it's "security by default" out of the box. If anything, I would think the government would err on the side of security (so their government hire doesn't get the bleeding edge driver or graphics utility, boohoo).
If the gov does use Linux widescale, I would think they should scrutinize all the distros, and come up with one STRONGLY suggested one so all machines will be compatible and fixes can be applied everywhere at once.
It's 10 PM. Do you know if you're un-American?
Well, okay, you can have Motif.
---------///----------
All generalizations are false.
--
I like to watch.
NT Cripples Navy Cruiser
134340: I am not a number. I am a free planet!