I have been so totally overwhelmed with projects that I have had very little time to read, research, or blog. So I was excited this morning to take a few minutes to download the new SDL research paper from Microsoft’s blog. It examines vendors using Microsoft’s SDL in both Microsoft and non-Microsoft environments. And what did I learn? Nothing. Apparently their research team has the same problem as the rest of us: no good metrics, and the best user stories get sanitized into oblivion. I am seriously disappointed – this type of research is sorely needed.

If you are new to secure software development programs and want to learn, I still encourage you to download the paper, which raises important topics with snippets of high-level information. As a bonus it includes an overview of Microsoft’s SDL. If you aren’t new to secure development, you would be better off learning about useful strategies from the BSIMM project. If you are a developer and want more detailed information on how to implement Microsoft’s SDL, use the blog and the web site. They offer a ton of useful information – you just have to dig a bit to find what you want.

Back to the subject at hand: There are two basic reasons to examine previous SDL implementations: tell me why I should do it, and how do I do it. Actually three if you count failure analysis, but that is an unpopular pastime. Let’s stick with the two core reasons.

Those who have built software with secure coding techniques and processes have seen the positive benefits. And in many cases they have seen that security can be effective without costing an arm and a leg. But objectively proving that is freaking hard. Plenty of people talk about business benefits, but few offer compelling proof. Upper management wants numbers or it’s not real. I have made the mistake of telling management peers, “We will be more secure if we do this, and we will save money in the long run as we avoid fixing broken stuff in the future, or buying bolt-on security products.” Invariably they ask “How secure?” “How much money?” or “How far into the future?” – all questions I am unable to answer. “Trust me” doesn’t work when asking for budget or trying to get a larger salary allocation for a person who has been trained in secure coding.

It is very hard to quantify the advantages until you are coding, or trying to fix broken code. One of the advantages at larger financial firms is that they have been building or integrating software for a long time, have been under attack from various types of fraudsters for a long time, and can apply lessons from failed – and poorly executed – projects to subsequent projects. They have bugs, they understand fraud rates, and they can use internal metrics to see what fixes work. Over the long term they can objectively measure whether specific process changes are making a difference. Microsoft has. This report should have. Developers and managers need research like this to justify secure software development.

So where do you start? How do you do it? You ask your friends, usually. The CISOs, developers, and DevOps teams I speak with use tools and techniques their peers tried and had good experiences with. You have the same problem as your buddy at BigCo, and he tried SDLC, and it worked. Ideal? No. Scientific? Hell, no. It’s the right course of action, for the wrong reasons. Still, though, peer encouragement is often how these efforts start. Word of mouth is how Agile development propagated. Will a company see the same successes as a peer? Almost assuredly not. Your people, your process, your code. Totally different. But overall, from a decade of experience doing this, I know it works. It’s not plug and play, there are growing pains, and it takes effort, but it works.

On to the Summary:

Webcasts, Podcasts, Outside Writing, and Conferences

Favorite Securosis Posts

Other Securosis Posts

Favorite Outside Posts

  • Adrian Lane: Crooks Hijack Retirement Funds Via SSA Portal. Great post and very informative regarding a growing problem with fraud. And the onus is not on every person with a social security number to fix the SSA’s operational problem – the SSA needs to a) do a better job vetting users, and b) stop payouts through pre-paid cards. That entire arrangement is an uncontrollable clusterfsck. They put the infrastructure on the Internet, so they are responsible for operational security. Not that it’s easy, but intractability is why many IT projects don’t get started in the first place.
  • USA Today interview with Jony Ive. Some tidbits on design, and the one I really like is the focus on making functions invisible. His example of Touch ID is perfect – it just works, no “scanning… AUTHENTICATED” animations.
  • Mike Rothman: Is the Perimeter Really Dead? Of course not. But it’s definitely changing. Decent take on the issue in Dark Reading.
  • David Mortman: Managing Secrets With Chef Vault.

Research Reports and Presentations

Top News and Posts

Blog Comment of the Week

It was impossible to differentiate legit comments from blog spam this week. We will try again next week.