By the numbers: Windows vs Linux security

John McCormick, TechRepublic

In TechRepublic discussions about the virtues of Linux vs. Windows, Linux enthusiasts love to point out that their OS is much more secure. They complain that Windows and its applications are full of bugs and poorly written code and that Windows is more susceptible to viruses.

Of course, administrators who work primarily with Windows take exception to these arguments. They believe that Windows can be secured just as well as any other operating system and that Windows is simply targeted more often because its use is so widespread.

One way to get to the bottom of this dispute is to look at the numbers. In this case, let's examine the statistics on discovered vulnerabilities tracked by Security Focus Bugtraq. One word of warning: These numbers may just surprise you.

Buqtraq vulnerabilities 2001

Table A shows a cumulative list of vulnerabilities discovered so far in 2001.




Buqtraq vulnerabilities 2000
Table B shows the 2000 vulnerabilities listed by the same source.



The bottom line
As these numbers illustrate, Windows NT 4.0 was the leader in bugs identified during 2000. But Linux was not far behind. And in 2001, Windows 2000 has stabilised a bit and is actually running in the middle of the pack. One logistical note: It wouldn't be fair to add those Linux bug numbers together--most are the same bugs across every platform. However, the conclusion here is that there is obviously a comparable number of security problems with the various flavours of Linux, as well as Sun's Solaris, as there are with Windows NT 4.0 and Windows 2000.

Ultimately, with the vast number of individuals and businesses using Microsoft software, any flaws in the Redmond product are magnified because of their sheer impact. All this doesn't mean that I don't like Linux or that I'm a champion of Bill Gates and his Microsoft cronies. But I believe that simply because of its vast market share, Microsoft should be feeling a tremendous responsibility to make certain that its software isn't just profitable. It should also be as secure as it can be made because any problems will have such a huge impact. Unfortunately, the Redmond giant doesn't appear to feel that responsibility.

Microsoft should continue to have its feet held to the fire when a mistake is found. Conversely, Linux bugs will continue to be minor news unless that OS gains a larger share of the world's computers and thus seriously impedes the work of many businesses and consumers.

TechRepublic is the online community and information resource for all IT professionals, from support staff to executives. We offer in-depth technical articles written for IT professionals by IT professionals. In addition to articles on everything from Windows to e-mail to fire walls, we offer IT industry analysis, downloads, management tips, discussion forums, and e-newsletters.

© 2001 TechRepublic, Inc.

Advertisement

Talkback 21 comments

    These statistics are meaningle ...Anonymous -- 02/10/01

    These statistics are meaningless without including the release dates of the software in question. A product that only had 1 month in the marketplace during 2000 will have less bugs discovered than a product that was there for the full 12 months.

    On a simlar line, it's not reasonable to be comparing a mature operating system with one that was released more recently. Of course a more recently released product will have more bugs.

    These stats need some dates and even market penetration (which is pretty difficult to assess) to make them usable figures for comparison.

    Guys, The numbers shown there ...Anonymous -- 03/10/01

    Guys,

    The numbers shown there are surprising. Would it be possible to overlay the severity of the vulnerability on this? For example how many of the vulnerabilities allowed full access to the target box? How many were DoS etc?

    Ben

    There are 3 major issues not a ...Anonymous -- 03/10/01

    There are 3 major issues not addressed:

    1) Roundtime for the vunerability to be fixed.
    2) Severity of the vunerability
    3) Source of the vunerability (code review vs live exploit).

    In the case of 1, most free-software vunerabilities are corrected very quickly once they become known due to the availability of source code, and the fact that its in everybodies interest that the vunerability is corrected ASAP.

    In the case of 2, not all vunerabilities are of severe consequence, but still get listed on BugTraq. One such example was a buffer overflow in the BitchX IRC Client which I submitted an analysis and patch for. This vunerability was minor - it could be exploited as a DoS, but did not affect the host system in the majority of cases.

    In the case of 3, vunerabilities found through code-review are moderately low risk - often they are purely theoretical, and have no live exploits in the wild. These are only possible because the source code is readily available for all to review. Closed source software vunerabilites are almost always trial by fire - somebody has compromised the service, and the exploit is in the wild.

    How many vunerabilities were p ...Ken Foskey -- 03/10/01

    How many vunerabilities were posted within Microsoft and not publicly posted? LInux is very public, Microsoft is very private.

    If you do anything web related ...Anonymous -- 04/10/01

    If you do anything web related stay clear of Microsoft or your system(s) will soon be compromised.

    most of the bugs in linux are ...balthazar gonzales -- 14/10/01

    most of the bugs in linux are discovered by either the creators of the software or by the people who use the software. and when bugs are found,they are immediately patched. no,if,ands or buts about it! another thing to point out is the actually patch development time is much,much higher with windows than linux...you might have to wait weeks before a patch is released.

    you also have to take in effect linux has 500% more software included with the os compared to win/nt/2000/95/98/me/3.1 has combined.

    also if you were to use linux with the base packages....you would have a really secure system.

    lol its about time someone poi ...Anonymous -- 14/10/01

    lol its about time someone pointed this out linux commies always think there OS is Unstoppable and better then any other which isnt the case anyone with the proper intension could exploit linux amazingly but since no ones getting anything out of screwing up 3% of the pc market why bother ? i dont want windows to be supperior just give it the same respect linux gets.. anyone who would want to take a shot at linux's security would get the same end result as an NT platform

    GOOOOOD DAY SR.

    you might have to wait weeks b ...Anonymous -- 14/10/01

    you might have to wait weeks before a patch is released.
    _____________________________________
    NO business has ever waited a week for a patch to fix a bug MS was already aware of ! stop the BS and there is no company for linux so if anything it takes linux longer to patch a linux problem red hat might fix a patch that has cropped up from a redhat version of linux but when its a widespread bug its not fixed as fast as you think..

    Quote.. If you do anything web ...Anonymous -- 14/10/01

    Quote..

    If you do anything web related stay clear of Microsoft or your system(s) will soon be compromised.

    ____________________________________________

    your an **** and you trick your self into thinking linux is superior and that 1 day itll run on every pc in the world but ya know what ? NOT GONNA HAPPEN! so stop trying to scare the less savy from windows , windows |> Linux.

    LOL DoS is dead linux luv4r th ...Anonymous -- 14/10/01

    LOL DoS is dead linux luv4r the charts show high risk bugs not bugs on things like OMG i typed the letter A and it didnt show up the first time..l00zer

    DoS is dead? hardly. With the ...Paul R. -- 15/10/01

    DoS is dead? hardly.
    With the release of Windows XP around the corner DoS attacks have only begun. With the implementation of raw sockets people will be able to send packets without any "return address", essentially making that OS a safe-haven for script kiddies. I respect windows 2000 as it is the superior windows operating system, but i can't say the same about xp.

    Gibson Research article explains it best
    http://grc.com/dos/winxp.htm

    Linux has its own problems to deal with, but you'll be seeing a lot more of it in the future due to it's open source nature and adaptability. At this point, however, linux is playing catch-up with windows. as soon as that ends, the real fun will begin, and it will be interesting to see what these two groups (windows programmers, linux programmers) come up with.

    Anonymous , if you are a netwo ...Dean -- 15/10/01

    Anonymous , if you are a networking engineer I am betting that network is made up of speak and spells.

    It's for the kiddies of the employee's , in the creche , and you shut down the network in the afternoon when the kid's ain't using them .....cause they are having their afternoon nap.

    This gives you time to correct any spelling they did on said speak and spells while you drool over the thought of going home and firing up your PC with pre-installed Windows Me to play MineSweeper.

    The OS's are used for differen ...Anonymous -- 15/10/01

    The OS's are used for different, I, for one, am tired of anti-MS linux enthusiasts. If it weren't for MS, we wouldn't have linux, why you ask, BECAUSE IBM WOULD BE DEAD.
    Linux, as it is now, is a pile of trash for home users, it cannot be taken commercially for an average user to use, how many people honestly know what theya re doing with computers? Not many, I know this from experience.
    However, on the business side, linux still can't be king, as it still lacks much of the functionality that Windows has, while windows lacks some things linux has.
    Yes, MS takes longer to patch things but think of all the compatibility issues microsoft has to deal with while linux has what? less than a quarter of the programs window has to deal with(thats being quite gracious its much less).
    If a security leak is found that is very dangerous its near instantly patched. And I'm willing to bet if Linux did get popular enough, virus writers and crackers would instantly use its open source against it. Gee wouldn't that be great, linux would be in over its head.

    OK I'm posting this message as ...Anonymous -- 18/10/01

    OK I'm posting this message as its clear to me that you are all overlooking a few certain facts. Firstly, who on earth would want to hack a windows box when they could have control over a linux box, which would have more destructive capabilities. Also I'm sure those figures will have rocketed since redworm virus and nimda. Finally though.. have you ever wondered why MS hasnt gone open source.. perhaps thier security might once agian be spat on.

    What made you choose the linux ...Anonymous -- 21/10/01

    What made you choose the linux distro's listed? Why not also try comparing slackware.

    Funny Logic. The real differen ...C. -- 22/10/01

    Funny Logic. The real different isn't in the amount of bugs found, but in the time the OS'es solved the problem. When a bug is found in linux the bug is rapidly fixed. When you find a bug in windows, you pay $90 to report the bug and wait at least a year in the hope they make something about it (IIS???).

    C.

    A superficial and juvenile ana ...Ewing Caldwell -- 07/11/01

    A superficial and juvenile analysis. The questions that need to be asked across ALL OSes, are:

    How many of those bugs were externally (remotely) exploitable?
    How many had a demonstrable remote exploit?
    How many of these are of any consequence? (is it a DoS? Is it a file system exposure? Is it a full, all privileges, login-type exposure?)
    How many are an internal exploit?
    Of the internal exploits, how many were only a theoretical possibility (eg: a user gaining root/admin privileges, a possible race condition in temporary file creation, poor file permissions losing a bit of privacy ... and so on)

    Each of these questions should be weighted according to the danger they represent. An incorrect file permissions problem for an application creating a temporary file is of no
    importance or consequence on a workstation used
    by only one person.On a multi-user server, however, it can have a completely different character: if it is the password file being modified by the system administrator, then it is
    rather more important than a publishing application's temporary scratch file, when everyone using the system is entitled to see it anyway.

    The questions to be asked for weighting these are: Is this of any consequence? If so, in what
    circumstances, and to what extent?

    For example:
    What does an internal forcing of root/admin privileges matter when it's a workstation used by only one person?
    What does it matter when it's a file & print server?

    What does a race condition matter within a single
    user workstation?
    Within a server?

    Just lining up the numbers is poor and lazy analysis. Each report should be counted, yes, but it should then be weighted for importance (danger from exposure, damages and so on). The weightings can then be added up and then you have it: the most dangerous OS or OSs to have and to hold.

    Perhaps one group of systems will comfortably outscore the other group should properly weighted scoring be done. Perhaps frighteningly so. But then, perhaps not.

    It is notable that ALL Microsoft's security problems have remote exploits, many of which either expose sensitive information or leave an admin empowered cmd.exe shell tied to
    a common service port.

    It is notable that "anti-hacking" insurance premiums are weighted against Microsoft's products with premiums for MS stuff costing up to 15% more to insure. Have the insurance companies done the sums? You can bet they have.
    So scare us. Do the analysis properly. Give system admins a REAL insight into what we are up against for all systems. Let's NOT play the "You've got more 'sploits than me" game. It's
    childish.

    How are these vulnerabilities ...Franck Routier -- 08/11/01

    How are these vulnerabilities counted ?

    I mean Linux distros include a LOT of sofware. What sofware are taken into consideration ? Same for Windows (IIS, ...?). If a problem is found, say, in Apache, is it counted once for each OS it applies to ?

    Just curious.

    no one seems to mention openbs ...Thomas -- 06/02/02

    no one seems to mention openbsd or any bsd here. linux is a poor mans bsd and i wouldnt use linux, solaris or windows for any critical servers, my choice has and always will be openbsd and freebsd for workstations.

    as for windows, i'll let reception be the help desk, "just reboot and you should be fine till the next BSOD."

    SECURITY FOCUS SAYS THIS IS BO ...Anonymous -- 10/02/02

    SECURITY FOCUS SAYS THIS IS BOGUS... I emailed the president of Secuity Focus and told him that they apparently need to do the WORK FOR THE IDIOT JOURNALISTS who seem to be totally unable to understand what the numbers quoted REALLY MEAN...
    He said he agreed and they are going to be issuing a regular report which spells out what these idiot hack journalists cant seem to grasp...

    They compare the totals for Windows NT to the number of Linux incidents x 10..

    Here is the lame logic:

    Windows NT incident (OMPARED TO) rsync on Redhat + rsync on Suse + rsync on Slackware + rsync on Redhat 6.1 + rsync on Mandrake etc..

    The problem is there are multiple vendors of Linux.. and each one issues a warning to their customers they are responsible to about a problem... and 99% of these issues are in a basic item that all linuxen have.. so while there is really a single incident about one security problem, say with SENDMAIL, it will be reported by all the different linuxes that are affected.. and these idiot journalists like to count that SAME FLAW 1000 times..

    Thats some seriously fu*ked up logic, and makes this article totally full of ****.

    That's a bunch of crap. First ...Anonymous -- 19/06/02

    That's a bunch of crap. First off, the bugs found in M$ are often major and impact on whole systems and will affect an intier LAN.

    *nix and the GNU community often find the bugs and reliese updats faster than the bad guys can respond and often are found before any bad guy found one.

    Microsofts attitude twords security is not their first concern. Where *nix it is the first.

    How many viruses are reported with root access? compare that number.

    I would also like to see the reported $ amount of damage caused to IIS vs Apache. Apache on *nix is about 60% of the servers, serving pages on the web. So that crap about MS is the domminent player here doesn't wash.

Add your opinion

Latest Videos

Sponsored content

Power Centre - Content from our premier sponsors

Blogs

  • Phil Dobbie Is wholesale-only backhaul just a pipedream?
    The potential acquisition of Pipe Networks by SP Telemedia has raised the question about whether vertically integrated backhaul providers will mean higher wholesale prices for ISP customers.
  • Array Get extensions going in Firefox, redux
    Previously on Null Pointer we looked at getting extensions working in Firefox betas, and that was great until the fine folks at Firefox changed their minds.
  • Array How reliable is IP telephony?
    Have you ever heard a weird kind of hissing, crackling or popping noise when calling someone on an IP telephony line? How rare is the phenomenon these days?
  • More blogs »

Tags

Back to top

Featured