But the most important post I will write in 2012 is this one because it asks a question that has haunted me all year, and I do not expect it to abate in 2013.
When looking at IAM programs, I recommend taking a Crawl-Walk-Run approach given the many strategic, tactical, and integration challenges to deal with in any IAM project. There's lots of new technology to deal with, new business processes (such as automating a formerly ad hoc, manual provisioning system), and teams who've never worked together now working in close collaboration. For all these reasons and plenty more, it makes sense to be conservative with what you can realistically achieve in IAM at each hop in the Crawl-Walk-Run cycle.
If you are just getting started, then carving out a set of use cases that you can count on one hand (or even one finger) is a good way to think about what you can realistically achieve with IAM. Start with something you can actually deliver on -- say, single sign on, or pick your favorite, and then build on this success. This avoids Battlestar Galactica project plans, deep seven-figure project spend of mega IAM suites, and, best of all, you will still have your job at the end of the project!
Following this basic plan is a proven low-risk, high-reward potential way to get IAM work done in most companies. However, it leaves open on critical question.
I was talking with a large company about just such a Crawl-Walk-Run strategy. We laid out the goals, architecture ideas, constraints, and progress tracking metrics. We came up with a reasonable plan to get started, and some ideas on direction for how to mature the program over the next few years. Then there was a long pause, and they asked a question that has no real answer in companies today: Who does this work? And how do other companies do this?
IAM projects get so wound up around tooling and processes that critical organizational questions go unanswered. It's striking to sit back and realize: This is not a one-off. It's the norm!
The answer to the question --bhow do other companies do this? -- is they cobble it together with some security people, a compliance person perhaps, maybe a part-time architect, and whatever developers are left laying around with spare cycles. In short, it's a hodge-podge.
This brings me back to my favorite Kent Beck quote: "I used to think of programs as things, but now I think of them as shadows of the communities that build them" Think about that statement in the context of your IAM initiatives. Is it a well-organized team with clear goals, or is it a slapdash effort? How do you think this will be reflected in the end product?
IAM tools, standards, and technologies made tremendous strides over the past decade. As an industry, we're getting pretty good at understanding how IAM processes should function; you don't see the "seven-figure IAM project gone bad, VP gets fired" fiasco as often these days.
So both technology and process are markedly improved, but what about people? Does your company have a VP for IAM? Or a director? The vast majority of companies have neither. Most do not even have an architect. Where does this leave us? Outsourcing, of course.
Outsourcing makes sense in some areas, and consultants have a role to play in IAM, but they should not be the full staff on these initiatives. IAM projects by their nature have distinct characteristics that require navigating multiple shades of gray in the security policy and its impact on usability. These subjective calls should not be outsourced.
Many companies launch IAM through security and compliance teams. They're all wonderful people, but their goals are often at odds with achieving broader adoption and maximizing value out of IAM.
IAM projects rarely offer a good return on their costs in the context of a given project. Most IAM solutions must be used by more than one project to be valuable, so ongoing visibility across projects is a critical success factor. Read this as: Your architecture team needs to lead.
So how should other companies do this? Some keys that we'll discuss in future posts:
- Realize that IAM is a megatrend. It's not going away. By their nature, distributed systems need dedicated and ongoing efforts to ensure security policy is enforced.
- Get real on IAM governance. If your company does not have an IAM VP or director, then you need to advocate for one. There is only so much progress to be made through the middle.
- Think about IAM as more than just compliance and security. It's a factor in every single mouse click a user makes. Widen the circle of IAM input to get broader and better quality adoption.
- Nominate a dedicated IAM architecture team who understands your business and enterprise architecture and can define a clear direction for IAM. In your company, follow ongoing industry trends and identify pragmatic ways to improve IAM across projects.
- Use consultants wisely -- as amplifiers as speed enhances, but not ends in and of themselves.
- Realize there are no silver bullets. Ensure your testing/QA team is able to thoroughly test all IAM efforts that emerge. Arm them with knowledge on how the protocols should work and test cases to ensure that they do.
The bottom line here for companies is to not think about IAM as a point-in-time effort. Effective IAM requires effective people -- executive buy-in, a mix of skills, and teams who can get the job done.
Gunnar Peterson is a Managing Principal at Arctec Group