In my last blog post, we looked at some of the hazards in identity and access management (IAM) projects. Specifically, most IAM project initiatives are created from a very dangerous place -- the auditor's spreadsheet.

Audit plays an important role, but blindly following an audit report into large-scale IAM "suite" (in name only) deployment, can result in seven-figure disappointment and hinder career upside.

IAM efforts should be architecture-centric, not audit-centric. Audit is an input to architecture, but IAM architecture focuses on assets, risk, standards, and integration -- how to deliver on the requirements in a real-world system, not simply completing the audit Checkbox Olympics.

The value of an architecture-centric approach to IAM is clear, but the next question becomes: What processes can be used to get there? Outside of specialty consulting groups, there are limited options for industry-standard IAM design, development, and deployment processes.

As IAM matures, this will likely change much the same way secure coding has been progressively baked into the Software Development Lifecycle. Enterprises now have a number of choices, such as BSIMM and Microsoft's SDL, that offer concrete ideas, guidance, and sequencing on improving security in code.

IAM faces similar challenges to those who drove the movement to BSIMM and SDL, but the IAM space lacks a "you go in A, you come out Z" cohesive approach to delivering IAM in real-world systems.

So while there is not a fully baked IAM process out and available, there are some starting points to consider. Taking an architecture-centric approach to IAM means starting with, wait for it, architecture. There are several sources to consider, including The Open Group Enterprise Security Architecture, which covers much of the core territory for IAM architects to design and plan for.

Delivering on architecture is the next pressing challenge, and here the work from BSIMM on software security is instructive. BSIMM's framework has four main categories: Governance, Intelligence, SSDL Touchpoints, and Deployment. Each of these has a direct implications for the IAM project life cycle:

In conclusion, BSIMM's focus on software security is not a one-for-one for IAM's top concerns; it's missing some areas, such as operations that are important for IAM efforts. BSIMM requires some adaptation to be useful in IAM, but most of the core bits required to get a successful IAM project up and running are present in BSIMM. Hybrids are the strongest plants in nature. If you're starting an IAM project now and need a framework to adapt for IAM, I would recommend starting with BSIMM.

Gunnar Peterson is a Managing Principal at Arctec Group