Reflections from INCOSE 2012 on Making ‘Just Enough’ Traceability Effective
AndyGurd 270001QKDH Visits (5070)
I was lucky enough last week to travel to the INCOSE (International Council on Systems Engineering) International Symposium 2012 near Rome, Italy. An excellent opportunity to meet the systems engineering community and hear about their interests and concerns. We had lots of traffic to the very stylish IBM booth where we talked about the IBM Rational solutions for systems engineering and the latest from IBM Research on tool interoperability and design optimization & trade-off. I’d like to claim the traffic was to due to my presence but in fact there was lots of excitement and interest in the must have giveaway of the conference, the IBM Limited Edition of Systems Engineering for Dummies book
(if you weren’t there and don’t have a copy, you can download a PDF version).
Being at the INCOSE event reminded me of the very active and interesting discussion I recently provoked on the INCOSE LinkedIn group with the posting of the link to my previous blog post ‘Traceability – How Much is Enough?’. It’s a great read with some very provocative statements about whether traceability is at all useful and that it’s the root cause of failure on projects that overrun and overspend versus those that say it’s absolutely vital on safety-critical systems or where the project is contract-driven. In the end I think some consensus was reached between these two camps that ‘just enough’ traceability to keep a project on track, provide customer/market need context to engineers, facilitate impact analysis, and (if needed) to meet industry standards and regulations, is sufficient. Any more is excessive and wasteful and likely to bog down progress towards to delivering innovative products and systems.
During a quiet time at the IBM booth, I also had chance to chat with my colleague Brian Nolan (marketing manager for aerospace & defense industry at IBM Rational) about effective traceability, since Brian is very interested in this topic and has presented on a Dr Dobbs webcast on ‘3 Ways to Improve Traceability and Impact Analysis’. Brian believes in what I would describe as ‘traceability by design’, meaning that traceability is automatically established while you decompose your system design (for example, use case to use case realization to sequence diagram and so on). This discussion also reminded me of what another colleague Greg Gorman (program director for IBM systems and software engineering solutions and the INCOSE Corporate Advisory Board member from IBM) described several years ago as ‘link while you think’, meaning traceability is created by the tools, while you are performing requirements decomposition, design and development, rather than as an overhead activity afterwards.
I think we’ve now moved some way beyond ‘link while you think’. While an information model with ‘just enough’ traceability for your project needs is essential to avoid traceability spiraling out of control, with new approaches such as Linked Lifecycle Data from the OSLC (Open Services for Lifecycle Collaboration) community, and tools that recognize implicit traceability, provide new ways to visualize lifecycle traceability and perform effective impact analysis, we can make traceability work for us to help engineering become more agile, while staying within costs and schedule and produce innovative, higher quality products and systems.