Bristol Group Local Event Review - 28th September 05



topbar.gif (2145 bytes)


Review of the Bristol Local Group event on 28th September 2005 - “The Tool Vendors’ Response to Architectural Frameworks”

Two presentations from the meeting are now available:

In June, Bristol Local Group held the first of two meetings on Architectural Frameworks in general and MoDAF in particular, with contributions from MoD and academia.  This, the second meeting, was the tool vendors’ turn to put their point of view.  Forty-six people were present to hear three excellent speakers giving their diverse views.

Fran Thom, from Artisan, showed how UML, SysML and MoDAF formed a natural progression, each being founded on the strengths and wide acceptance of the previous. The strength of a tool is to make systems expressed in MoDAF form readable and usable by a wider audience than those who fully understand UML.  This can be achieved by tailoring the language and the visibility of information to the type of user.

Toby Sumpter, from The Salamander Organisation, warned about the potential divisiveness of the MoDAF views if individual views are owned by different parts of MoD – as their names suggest will happen.  The tools need to be able to present a holistic view, being interpretable by soldiers as well as technicians to give a common understanding of the system and its operation.

Martin Owen, from Telelogic, showed how requirements and design needed to work together, albeit not necessarily supported in the same tool but in an integrated suite.  In this way both high and low level design could be appropriately accommodated, both driven from linked requirement sets.

As there hadn’t been much time for questions at the first meeting, Dave Mawby and Ian Bailey, from the MoDAF team, and Rick Adcock, from Cranfield University, joined the three presenters to answer questions from the audience, summarised below.

  1. Where are we today with MoDAF?  The Internet version of Issue 1 has been placed on the web today.
  2. Are there frameworks for use in commercial areas where there is no coordinated customer?  Architecture frameworks did not arise in defence (eg Zachman) and are really only a way of presenting the business rules of whatever industry you are in.
  3. Is there a security view?  Whilst there are some in TOGAF (The Open Group Architecture Framework), there isn’t one in MoDAF.  Security will percolate the whole model, possibly as attributes of information, rather than be confined to a separate view.
  4. Is the static MoDAF analysis suitable for the fast-moving battle-space, where the soldier taking initiative causes the unexpected?  It is possible to build dynamics that bridge between static states.
  5. Do the different models in the Telelogic scheme force consistency, as a single model would?  The various models are built on a common data model (or meta-model) in different data repositories.  There are active links between them.
  6. MoD personnel change posts frequently; will MoDAF-related work need to be undertaken by outside contractors?  There is already a lot of outsourcing and, yes, it is difficult to get new skills into the MoD.  It is hoped that MoDAF will assist MoD personnel to express what they want in a consistent way, rather than them necessarily being skilled in the capture process. The source of information would need to be captured, but the source would not need to be the one to capture it.
  7. The tendency for people to shy away from learning the models will mean that purchasers of complex systems will not know what they are buying and, hence, will buy the wrong thing.  Already, by the policy of moving people from one post to another (possibly unrelated) one, purchasers may lack domain knowledge or purchasing skills.  The introduction of MoDAF will not worsen this.
  8. If each organisation has the freedom to choose what they capture about a system, how will this benefit integration of independently acquired systems?  Exemplar guides will be produced to show how to achieve particular tasks; this should encourage consistency.  However, one view was that a project cannot be required to capture more information than is currently useful to it.  This was in contrast to another view that there may have to be certain mandated information recorded.  “We have thrown you a dictionary, now you have to learn how to write novels”
  9. What about Configuration Management?  This is still a massive problem, currently without taxonomy.
  10. Is there cooperation between tool vendors?  MoDAF mandates how to store information, so one tool can put data into the repository and another take it out.  Vendors will compete on the usability of their tools, not on the information they will produce.

 

 

topbar.gif (2145 bytes)

Last Updated: 09 March, 2006