If you haven't already, take the time to read Mathew J. Schwartz's piece, Secure Coding Or Bust. The column provides an interesting overview as to why secure software development is important.
It also provides a few good suggestions for a start, such as the Software Assurance Maturity Model (SAMM), the Building Security In Maturity Model (BSIMM), and Microsoft's Security Development Lifecycle.
These are all excellent, but don't include some of the most important hurdles that need to be leaped before the program can get running at speed. We will dig into three of those below. But first I'd like to draw your attention to an important technical resource, the Open Web Application Security Project, or OWASP. It's a vibrant security application development community that provides security how to guides, information on common threat vectors, attack techniques, and insight on most types of vulnerabilities that plague web applications. There's plenty to consume there. I suggest you dig in.
Now, onto the three essentials you'll need to win:
Get A Champion
Getting any application security program off the ground has as much to do with garnering executive backing than technical and application security prowess. Why? Because application security affects everything about your development program, from how much it costs to how defects are handled to developer training, and what developers are asked by the business to prioritize; it can even slow release times. Building application security into a program where it didn't exist before isn't easy.
What to do about it? Get an executive champion. Someone who is high enough up on the organization chart to provide the political air cover when things get tough. And they'll get tough especially when developers are being pressed to move an application forward to production rapidly--even when it has critical vulnerabilities that will "be dealt with later." Yes. That's when the knife fights start. Frankly, many--if not most--security teams don't have the power to slow down development times to address security concerns. They can advise, but they can't always make it so. An executive with this authority to slow development--when it's needed--is essential during certain times. And they'll help you with everything else you won't find in a security framework: such as getting the budget you need and convincing others that secure application development is in the best interest of the company.
Trust me: most executives don't get this stuff. And you'll need someone who has the power and the ability to fight your fight in the corner office.
Enlist The QA Teams
Having an executive champion for a secure application development program is also important to help change the way development functions. Now, rather than barging in on developers and declaring how everything they've coded until now has sucked and "you are about to show them how it is done right." You might want to try a different tact. That's could be to convince the organization that security defects should be treated and remedied as part of the normal Quality Assurance (QA) process. The reality is that many organizations don't treat security defects in the same way they treat software defects that affect performance and availability. However, making security defect vetting part of the normal QA process will go a long way to steer development teams on the right track than the security group acting like the new vulnerability sheriff is in town.
Fight The Urge To Do It All From The Beginning
Once you start looking for them, you are going to find more application vulnerabilities than can be possibly dealt with. Like flipping the light switch on a kitchen that was left neglected for awhile: the roaches and rats will be all over. Only when you illuminate your application security program it won't be insects and rodents you uncover, rather it will be more than a dozen of vulnerability classes such as cross-site scripting, buffer overflows, SQL injection, and username enumeration that occur over and over again in applications. There are probably too many vulnerabilities, depending on the size of your organization, in existing software to go back and fix every application. So start smart with a short list of common vulnerabilities you're finding in newly developed application. Start fixing those first. Slowly build from there by adding new vulnerabilities to the list that QA teams look for during their process.
You won't solve every software security problem with this, but it's certainly a leap in the right direction. And not as easy as downloading a secure application development framework and running with it.