Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Developing Securely In Windows

Posted by samzenpus on Mon Nov 21, 2005 01:47 PM
from the the-secure-collective dept.
FrazzledDad writes "No, really. Please pick yourself up off the floor and stop laughing. Yes, there are good books on developing Windows software in a secure fashion. Keith Brown's The .NET Developer's Guide to Windows Security is right alongside Howard and LeBlanc's Writing Secure Code as examples of good Windows security works. Brown's book should be on any .NET Developer's bookshelf and will be of use to developers who work in other development platforms on Windows." Read on for the rest of the review.


I know the entire topic of Windows security may kick off a "slightly" enthusiastic debate among Slashdotters. I'd really prefer not to get wrapped up in a fray, so let me just say that a professional software developer needs to well understand the security issues in the environment and platform they're working on. This book's an important aid in that understanding. Great Fundamentals

Brown's book is broken into six parts, ranging from "The Big Picture", an overview of security on Windows, to "Access Control" and a wrap-up "Miscellaneous." Each part is made up of numerous "items," one topic which Brown elaborates on.

Brown covers a lot of very basic, important fundamentals such as "What is Authentication?", "What is a Luring Attack?", and "What is Kerberos?" He gives concise, clear overviews of each topic, then gets into the weeds where necessary.

For example, one of Brown's first emphatic points is that development on Windows platform shouldn't be done using an account with Administrator privileges. He covers the "why" in several early items, then spends 11 pages in Item 9 showing the approaches, tools, and issues involved in developing under a non-Admin account. This particular item needs to be stapled to far too many developers' foreheads because they don't understand, or care about, the ramifications of development as an Admin. Great Details

Brown also goes into great detail on many Items. His discussion of IPSEC is a good example. He spends Item 68 on the fundamentals of IPSEC such as key exchange and authentication, then goes on in Item 69 to discuss the details of implementing IPSEC via policies in a domain. He covers client and server configurations, then gives rationale for selecting various options. He also talks about why it's not the best solution, or even a complete solution, but does point out where IPSEC makes sense.

COM programming gets an entire section/part to itself, and Brown does a great job explaining the complex issues surrounding securing COM(+) communication. He discusses Authentication, Impersonation, and what calls you need to make in your Main method to properly invoke various COM security aspects.

Threat Modeling gets its own Item, but isn't covered in great depth. Brown lays out Microsoft's STRIDE system (Spoofing, Tampering, Repudiation, Information disclosure, Denial of service, and Elevation of privilege) as a guideline for threat modeling. He also talks a bit about attack trees. Neither topic gets substantial treatment; however, Brown makes it clear he's only introducing these topics and points readers to several other resources such as Swiderski and Snyder's Threat Modeling. Great List of Cons and Problems

Part of good software engineering is understanding the ramifications of choices you make. Brown's very good about laying out the "Why" for his items, plus he's also clear where hard choices have to be made.

For example, in his discussion of IPSEC he asks "Where is IPSEC useful? When you don't have any better alternatives." He goes on to show how IPSEC can be used to help COM servers talk securely, or in .NET Remoting under the 1.1 Framework which stupidly doesn't provide secure communication channels.

Another example might be the erasability of a secret under .NET. Managed environments such as .NET and Java don't make it easy to ensure secrets (passwords, keys, etc.) can be erased out of the managed memory heap or at least overwritten immediately after their purpose is fulfilled. Not only can the object's memory be left unerased, but what about controlling whether it's written out to a swapfile? Brown points out these sorts of issues and tries to point out how to deal with them. What the Book Doesn't Cover

Brown's book isn't so much about specific coding techniques, although there are a fair number of those within. You won't find specifics on .NET's code access security, or issues around cross-site scripting. You'll need to look to Howard and LeBlanc's Writing Secure Code for code specifics.

Rather, the book is more about approaches to secure development on Windows. Brown's book also isn't about security and threat analysis, but again, he's forthright about that and points readers to other sources.

Bill Wagner, author of Effective C#, points out on his blog that Brown's book would be more usable if "titles [were] organized around the tasks I need to perform." I think that's a good criticism - a cookbook format would be a great improvement for a second edition. Summary

The book's very well written with a good index and a terrific Bibliography which serves as a great reading list for furthering one's knowledge of security on the Windows platform.

I've found the book very educational and useful. It's an important addition to my bookshelf and has already helped me with a couple of important topics. I think any professional, contentious developer working in the Windows environment would find this a vital addition to their bookshelf as well."


You can purchase The .NET Developer's Guide to Windows Security from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
+ -
story
This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • by PeeAitchPee (712652) on Monday November 21 2005, @01:50PM (#14083676)

    From Amazon.com:

    Customers who bought this title also bought:

    • My Plan To Catch The Real Killers by O. J. Simpson
    • My Little Book of Marital Ethics by Bill Clinton and Rev. Jesse Jackson
    • 101 Iraqi Exit Strategies by G. W. Bush
  • ...next to "Building Castles on Sand".
  • Here's a thought (Score:5, Insightful)

    by sexyrexy (793497) on Monday November 21 2005, @01:52PM (#14083706)
    Instead of bashing MS and Windows, prove that you're the better programmer by compensating for the sometimes flawed security. If you don't think users can trust their OS, at least you can take pride in the fact that they can trust your software running on it. A solid piece of software is just as impenetrable on Windows as it is on Linux or any other platform - it's all about understanding the environment. Looks like a great book, thanks for the review.
    • Security isn't something you can just dump in to your program. It's a mindset you use to build a good program. Problem is if you build a brick wall on quicksand, your efforts are nearly useless. Compensation is not achievable.
      • Exactly. I mean its not like you could build a brick box that would provide boyant for the weight of wall and then build the wall on top of the box. I think the analogy you are looking for is "It's like building a battleship on the ocean".
      • And if you build a program that is running on windows, but behind a firewall that only exposes the port your application runs on... your application is the only attack vector for that machine... Yes, windows has flaws, mostly on end user machines, because of software that is poorly written to not install or run properly in a reduced context.

        That doesn't mean it should not be considered... Well, people are going to get drunk, and drive anyway, should we just eliminate the laws making it illegal?
        • God, I hope some developers are reading the parent. I am a Windows Admin for a smallish group (~450 machines) and almost all of the problems we admins run into result from applications doing dumb things. We have finally gotten a policy through our business admins that allows us to nix software that won't run properly as a user in the Users group, and we have had only 2 security incidents since (and one of those was a machine we didn't know existed and thus wasn't being patched or getting AV updates). Ther
        • And if you build a program that is running on windows, but behind a firewall that only exposes the port your application runs on... your application is the only attack vector for that machine...

          By the time I finished reading your post I thought of two attack vectors you neglected.

          1) The TCP/IP implementation (usually implemented in the....Operating System!)

          2) The firewall implementation.

          I'm sure if you give it ten minutes thought you can come up with many others.

          The O/S is important.
          • Well, the tcp/ip stack for windows is based on bsd iirc.. beyond this, I never said the firewall was software and ran on the windows machine... for that matter, there's also router attacks, and other types of DOS attacks.

            I didn't mean to infer that there weren't possibly other means of attack, so much as that they are much less a possibility... I'm not aware of any attacks exploiting the windows tcp/ip stack directly...
            • Well, the tcp/ip stack for windows is based on bsd iirc
              I'm afraid you recall incorrectly. Read this slashdot thread [slashdot.org] to get more info - but essentially, whilst there is plenty of BSD code in microsoft products, its unlikely that they used a bsd based tcp/ip stack in windows versions beyond NT3.5.

              beyond this, I never said the firewall was software and ran on the windows machine...

              Where they are running is irrelevant to this discussion.

              for that matter, there's also router attacks, and other types of DOS att
    • by aztracker1 (702135) on Monday November 21 2005, @02:04PM (#14083810) Homepage
      Io be honest, I think that isn't entirely true, there are many different vectors of attack.. the best we (as developers) can do is make sure it isn't because of *our* software. We can check input before processing, we can use proper error handling to prevent a default error handler from exposing a dbms password. We cannot fix windows services, or software we rely on...

      However, I have to submit that most of the time security exploits in windows, are user initiated. Reduced privileges for default users would be a great improvement.. aside from that, we must bear personal responsibility for what we create.
    • Definitely true, writing secure apps goes a long way. I think two of the largest contributors to Windows insecurity are a) Running everything as Administrator; and b) Clueless configuration (They do have some relation).

      Someone who has a mediocre grasp on system configuration can lock down a Windows box pretty easily. When we think of Linux this is a given because for the most part clueless users wouldn't be using Linux to being with. Though if I just threw up some Red Hat 7.x box right now on the net

    • Re:Here's a thought (Score:4, Informative)

      by tpgp (48001) on Monday November 21 2005, @02:09PM (#14083847) Homepage
      A solid piece of software is just as impenetrable on Windows as it is on Linux or any other platform

      Hmmmmn....How about if I change your statement to:

      A solid piece of software is just as impenetrable on Windows 95 as it is on Windows XP Professional or any other platform

      This statement is incorrect - Windows 9x Filesystem (for starters) makes your program's config files (or hell binary) overwriteable by anyone in the system.

      So you see - the OS is important.
    • Yes, that makes sense. If you leave your front door unlocked, at least make sure that your windows (no pun intended) are secure - it'll make you feel so much better when the thieves enter your house through the front door and steal all your stuff.
    • Like someone said, writing a secure app for Windows is like putting a padlock on a paper bag. Aren't the biggest Windows Security risks IE and Outlook anyway? With their html enabled embedded shiny gewgaw features to bedazzle the clueless?
    • Instead of bashing MS and Windows, prove that you're the better programmer by compensating for the sometimes flawed security.

      My modpoints ran out just as I was about to bump you up, but then, it looks like you're sufficiently well bumped by now anyway.

      No platform is or will be 100% secure, but that doesn't and shouldn't stop engineers from reducing the number of security holes.
    • A solid piece of software is just as impenetrable on Windows as it is on Linux or any other platform

      That's only true to a point. It's like having a VERY sturdy roof on a framed house with holes in your studs.
    • Interestingly, as part of my Computer Science degree at the University of Leeds I take a Secure Computing module in my final year, sponsored (ironically) by Microsoft.

      http://www.microsoft.com/uk/press/content/presscen tre/releases/2003/03/PR03013.asp [microsoft.com]
        • Tell the Army Corp of Engineers you can't build a foundation on sand. You'll get pwned pretty harshly.

          After getting pwned, give them the precision that you're not allowed to actually put anything INTO the sand, and watch them try to design some kind of floatable device on top of which you may be able to build your fortress for some time before they burst in tears.

  • by RapidEye (322253) on Monday November 21 2005, @01:53PM (#14083711) Homepage
    Brown's book should be on any .NET Developer's bookshelf

    Just putting it on the bookshelf won't help - ya need to read it before you put it there.

    On a related note, this pretty much rules out all the developers I work with - if you can't get it as a Book-On-iPod, they aren't interested these days =-)
  • Anyone else see the irony in having two articles on the front page of /. one of which is about secure coding and the other which is about "How To Write Unmaintainable Code"

    Or maybe making it unmaintainable is just a sneaky way of developing securely...
    the /. editors are sneakier than i gave them credit for
  • by Anonymous Coward on Monday November 21 2005, @01:56PM (#14083734)
    Can you point me to any OS where security is automagic and the developer doesn't need to consider it? Windows may have its problems and more than enough shitty architecture, but security must be considered in the design of Mac and Unix/Linux programs as well.
  • by LunaticTippy (872397) on Monday November 21 2005, @01:56PM (#14083747)
    I'm going to read this book.

    I am forced to develop windows applications from time to time, and I am ashamed of the poor security of many of these apps. It is definitely something that should be addressed. There are hundreds of thousands of part-time windows developers who know even less (gasp!) than I do about security.

    /me gets back to hard-coding plaintext database passwords.

  • by aztracker1 (702135) on Monday November 21 2005, @01:57PM (#14083752) Homepage
    Managed environments like .Net, Java, Python, etc. are a good start, but aren't the end of things. Most principles in security are similar regardless of platform. I do think a lot of the issues in windows security come from poorly written software that doesn't work properly in reduced security contexts... It's nearly impossible to develop on windows without administrative access to the machine you're working on.

    At least with a managed environment, you have less chance of attack channels which result from unchecked data. You still have to consider how software is used, what it connects to, and check data going to an rdbms in environments that don't do it for you (parameterized queries for ADO.Net are pretty nice).
  • Chapters 4-6 deal with setting up VMware on linux?
  • by Tibor the Hun (143056) on Monday November 21 2005, @02:00PM (#14083773)
    CWS: Hello secure app, I'm coolwebsearch!
    SA: Hi, I'm busy.
    CWS: HEEY! Look at me!
    SA:Uh-huh.
    CWS:You sure are secure aren't you?
    SA: Sure am.
    CWS: Hey, let's see which one of us is more important!
    SA: Whatever.
    CWS: MR PROCESS MANAGER!!! WHICH ONE OF US IS MORE IMPORTANT???
    Windows PM: It looks like you're trying to type a letter...
    CWS: YAAY! You know what,this sucks, I'm just gonna take all of the CPU cycles and all the network bandwidth now!
    SA:...!!...No..Must...keep......working..m....u... ....sT......
    Windows PM: Well, it looks.... like you guys..... are busy, I'm gonna take a n......ap. Wake me up if anyone starts writing............... letters.kbyethx... ...
    BSD: "Pleased to meet you, Hope you've guessed my name..."

  • A chain is as strong as its weakest link.

    The point is making sure your app is NOT the weakest link here, i.e. allowing a virus to inject code thru a buffer overflow or something. And we're living in post-Sony times, it was bad enough with viruses - just wait till the rootkits start spreading.

  • "He also talks a bit about attack trees." but doesn't discuss them thoroughly. I wish somebody would. I *really* wish there were a Web site devoted to them--something like design patterns. Perhaps powered by a Wiki.

    Where they're even known about, say by people who have read Secrets and Lies (Schneir 2000), far too many developers are forced to reinvent the wheel. This hurts security in a fundamental way. All too often, they've never been heard of. I'm glad they're getting some mention, on any level.
  • No, really (Score:4, Insightful)

    by Junky191 (549088) on Monday November 21 2005, @02:04PM (#14083804)
    "No, really. Please pick yourself up off the floor and stop laughing. Yes, there are good books on developing Windows software in a secure fashion." With snippets like that starting off an article, you start to see why people have trouble taking Slashdot seriously. This is getting ridiclous, almost FOXNEWSesque.
    • Re:No, really (Score:3, Interesting)

      With snippets like that starting off an article, you start to see why people have trouble taking Slashdot seriously.

      What, people won't take /. seriously if its contributors exhibit a keen grasp of the obvious?

      -jcr

  • Hate to do this but (Score:3, Interesting)

    by Anonymous Coward on Monday November 21 2005, @02:11PM (#14083861)
    this book can be read online for FREE as in beer or something. If you want it in one document you may have to get your "copy and paste" on, or if you are in hacker fever you could screen scrape it. Anyway http://pluralsight.com/wiki/default.aspx/Keith.Gui deBook.HomePage [pluralsight.com] yep all there for your Windows security mokery.

    Remember this is to build secure software on Windows, something that should not be frowned upon even if those who write Windows don't listen to this advice. So when your next Window app breaks and your customer is irate, you can say "uh uh that's MS Slammer 5002, that's a bug with Windows not my code buddy!! I know my shit and that's why you're paying me too much to do this, now stop bugging me already, don't you accountants do anything but make cups of coffee all day!!!!"

    Read the Book.
  • Developing secure software should always be platform independent. But security for developing an application does not stop at the developer, it continues through to the system administrators and the users. We've all heard software is only as secure as the O/S its running on, well its also as secure as the administrators/users running it. Security should be practiced at all levels from development to roll out and installation and through its entire lifetime. No software is completely 100% secure, so if f
  • Free Online (Score:5, Informative)

    by enkafan (604078) on Monday November 21 2005, @02:17PM (#14083920)
    The book was developed online via a Wiki, available here [pluralsight.com] for free. This is a great book that every windows and .NET developer should be aware of.
  • SELECT * FROM Windows WHERE security > 0
    fatal error segmentation fault
  • by digitaldc (879047) * on Monday November 21 2005, @02:21PM (#14083951)
    "Threat Modeling gets its own Item, but isn't covered in great depth. Brown lays out Microsoft's STRIDE system (Spoofing, Tampering, Repudiation, Information disclosure, Denial of service, and Elevation of privilege) as a guideline for threat modeling. He also talks a bit about attack trees."

    Didn't Gandalf use the assistance of Attack Trees to topple Isengard? Sarumon really needs to read this book.

    But seriously folks....another good reference here:
    http://www.microsoft.com/MSPress/books/5957.asp [microsoft.com]
  • by wardk (3037) on Monday November 21 2005, @02:28PM (#14084017) Journal
    Step one: Unplug from network
    Step two: Shut down
    Step three: Use power button to kill hung shutdown
    Step four: declare victory! you have successfully secured windows
    • You forgot to try to play the Final Fantasy victory theme as part of step 4, and suddenly realize you just turned off the house's only audio player*... ;)

      *for smarter people, this might not apply

  • by j1mmy (43634) on Monday November 21 2005, @02:56PM (#14084281) Journal
    int main( int argc, char** argv )
    {
        return 0;
    }
    • Right you are. But the software is only as secure as the OS. And if the OS gets 0wn3d, then the security of the application doesn't matter. However, I'll also say that the abilities of the administrator of the Windows box also have a lot to do with it. I know that I could run an Apache web server on a Windows 2003 server and never get cracked.
    • There's Firefox and Apache for Windows, isn't there?

      That would be so much funnier if IIS 6 didn't have so much better of a security track record than Apache has over the same period.
    • Windows has come a long way in the last 5/6 years and vista should ship reasonably secure out of the box. It's still an unbelievably shitty OS peddled by an objectionable bunch of borderline criminals, but you have to give them credit for addressing security issues.

      Isn't that rather like giving credit to someone who quits mugging, when credit should be given to those who have never mugged in the first place?

      • Yes, and that's a very logical behaviour when the former is currently the head of the state treasury. (Or the country president, or whatever you want that make him mugging being kind of a pain).
    • If the underlaying closed source API has bugs there's nothing much you can do about it.

      Does this mean that any system with bugs is not worth developing for? And if this is true does it mean that developers who code third party software should be absolved of any wrong doing? That's just nonsense to me. MS may not have a great product out there on the terms of security but the kind of retards at AOL who are coding weakly and creating a number of the exploits being used certainly are at fault.

      If we hold ou