20110109 arb sydney wgm minutes

Architecture review Board Meeting Minutes

January 9, 2011 

DRAFT

Preliminary attendees:Lloyd, John Koisch, Gary Dickenson, Ron Parker, Andy Bond, Cecil Lynch, Jane Curry, Steve Hufnagel, Roy Markami?, Grahame Grieve, Charlie Mead, Sorin (NITA), (check sign-up), Mead Walker

=Agenda(s)

1. Q1
   1. Call to order
   2. Approval of Agenda for quarter
   3. Approval of agenda for wgm
   4. Approval of Minutes
   5. TSC update
   6. Set Direction
   7. Solution to SAIF book stalloing.
   8. Meta-models from NCI(Charlie Mead/Sorin)
      1. Eccf
      2. BF
      3. GF
   9. Insight from DOD(Steven Hufnagel)
   10. Insight from Patrick Loyd
2. Q2
   1. Call to order
   2. Agenda
   3. Theoretical Frameworks
   4. NCI governance
   5. GF framework w/o implications
3. Q3
   1. Governance Operalization to present to TSC
   2. Adopton of DMP

Minutes

Q1

1. introductions (all)
2. Welcome to Arb (Ron Parker)
3. Agenda for WGM
4. TSC report (Charlie Mead)
   1. Austin has assumed chair of TSC. Austin is pushing for SAIF implementation guide apart from the SAIF book. The Enterprise Architecture Implementation has not solved cross-project management. Only way to overcome inadequacies is to deal with Governance. NCI has governance, not be-all end-all, which Charlie will present to Arb, to bless/modify for HL7, to present to TSC.
   2. There are two substantial discussions:
      1. Tooling - what do we need to support SAIF-IG?
      2. Consideration by TSC to generate another brand in HL7, an incubator thread: allows selected projects to rapidly develop implementations, not DSTU, taking existing work and making work - e.g. un-constrained RIM Compliant CDA. Would include governance.
   3. Ron Parker:Chuck gave TSC a tour through the business model, which has been blessed by the HL7 Board. Should value and manage the intelectual property(IP) of HL7. It is still in evolution. SAIF and ArB feature prominantly in the valued IP.
   4. You will see product descriptions floating around. There are 47 products.
   5. PCAST - There was a report issued by Presidents Council Advisors on Science and Technology. There has been a lot of flurry of response - ONC will respond. NCI wants to respond, but cannot directly. HL7 will respond, this week. NCI has adopted Thomas Earl's methodology, so NCI asked him to comment.
5. set direction for week:
   1. Monday
      1. Q1 SAIF tutorial
      2. Q2 SAIF tutorial
      3. Q3 Open
      4. Q4 Joint with EHR at EHR
   2. Tuesday
      1. Q1 SAIF Implementation tutorial(Ron Parker)
      2. Q2 SAIF Implementation tutorial (Ron Parker)
      3. Q2 Joint with InC at InC
         1. InC assuming ECCF
      4. Q3 Joint with MnM at MnM
         1. IG discussion
         2. Why do we need HDF instead of TOGAF? Ron Parker: TOGAF alone is not sufficient. Steve Hufnafel: Development Methodolgy of Togaf would provide it. Ron Parker: HDF is not complete. MnM/ArB need to dialog.
      5. Q4 Meeting Information Framework (Stephen Hufnagel, Cecil Lynch)
   3. Wednesday
      1. Q1 Joint with SOA at SOA
      2. Q2 Open
      3. Q3 Joint with OO/MNM at SOA
      4. Q4 Open
   4. Thursday
      1. Q1 Open
      2. Q2 Open
      3. Q3
         1. Approval of Agenda for quarter
         2. Planning for next WGM
      4. Q4
         1. Approval of Agenda for quarter
         2. Wrap up.

1. 1. SAIF book Stalling (Jane Curry)
      1. HL7 funded a SAIF book editor Karen Smith for producing SAIF in DITA. Karen trained HL7 staff on the procedures. ECCF and Intro have been converted to DITA. BF is in DITA but has not been updated. GF and IF are not complete. GF is being authored in DITA. DITA suggests use of concept maps. Persons comfortable with Cmaps will find it easy to use.
      2. Ron Parker: Use of DITA avoids reference problems between documents.
      3. Jane Curry: Each concept is a separate document, then the dita maps are used to assemble/link topics.
      4. Ron Parker: Premise is ability to version concepts, and be agile, maintaining coherence. Getting content put together is an issue. Concept mapping is a requirement.
      5. Jane Curry: Tooling with DITA is not the issue - hardest part is understanding what parts of document needs to be its own topic. DITA is an evolving standard. It does have commercially available tools. DITA was done as a pilot, and has not been approved for use with other projects. No decisions has been made to use further. There is a project to output from CDA template designer for May Ballot.
      6. Steven Hufnagel: It is too abstract.
      7. Patrick Loyd: SAIF and Sound is working on that.
      8. Ron Parker: The SAIF book is not for general consumption.
      9. John Koisch: Part of the problem with DITA/BF was the way DITA rendered the topics.
      10. Ron Parker: How are we going to solve the issue of how it is presented from DITA. We need to provide guidance to HL7.
   2. Meta-models from NCI(Charlie Mead/Sorin) Goal is to move from abstract description to usability.
      1. Eccf
      2. BF
      3. GF
   3. Insight from DOD(Steven Hufnagel)
      1. Currently VA/DOD are under mandate from Whitehouse to have joint EHR system. Have to make sure all of SAIF concepts are applicable. SAIF is an organizing paradigm, at least ECCF, allows to present assumptions. ECCF is nice scorecard. You can develop plan using ECCF to show where you are at. We are in a pbolem space where all I have to give is the Executive Summary. If you give people concepts of SAIF, they want documents on how to use it.
      2. Ron Parker: Concerned about lack of full document.
      3. Steven Hufnagel: HL7 is loosing control of SAIF - NCI, DOD, Infoway are each developing their own. ###Ron Parker:Solving for broad access representation is not a priority of the current authors. We(HL7) need an operational method - it is not the problem for the organizations doing the work.
      4. Jane Curry: The support for this approach is not trivial. To do the job requires someone with good english language skills, to serve as a resource to authors.
   4. Insight from Patrick Loyd
      1. Big issue is resourcing of highly-qualified authors - there are few who have the time, since it is often done by stealing time.
      2. Jane Curry: Karen was a good editor, but had no architectural skills.

Q2 Governance

1. Sunday TSC
   1. ArB will present to TSC our concept of governance. NCI has a concept of what governance looks like. It is a concrete plan. Charlie will present to Arb to discuss.
   2. Open group has paper on governance, so does Thomas Earl.
   3. We started with the open group material, then got proofs from Thomas Earl which was more complete.
   4. The problem was, that we did not find it finite enough. We took service development life-cycle and applied Thomas Earl's framework.
   5. We need to present similar to HL7. It is NOT service specific. We need to overlay governance over the life-cycle, applied to RISC.
   6. There have been quality initiatives. We need to apply SAIF/GF/RISK to address quality.
   7. NCI:
      1. There is a difference between business and architectural governance.
      2. Ensure compliance with ECCF
      3. Provide guidance for all lifecycle stages
         1. Apply to life-cycle of messages, documents, services.
   8. Open Group says you must have
      1. Compliance/Conformance
      2. Dispensation/Appeals
         1. Formally defined process
      3. Communication
         1. Educate communicate, and support governance
         2. Ensure govenance is understood and adknowledged by governed.
      4. Vitality
         1. Maintan the applicabiliy of the governance model
         2. Requires mode to stay curren by refining processes to ensure oncoing usage and relevance.
   9. Question - what is intersection between planning and budgeting.
   10. Thomas Earl
       1. Preceps, rules for decision making
       2. people, make decisions within constraints of precepts
       3. processes, coordinae decision-making activities, means and opportunities to control decisions, enfor policies
       4. metrics
   11. WOrking organizational structure

alt text

Charlie presented this structure from NCI.

1. 1. CaBIg/MD anderson
   2. MdAnderson wants answer on how to interoperate.
   3. If they wanted services in the repository, they have to follow the rules.
   4. CaBig can get service up/running in one hour.
      1. DOD has three governance boards
         1. Clinical - requirements for clinical
         2. Architecture - requiremenst for architecture
         3. Investment portfolio - budget
   5. Andy Bond: So how do you transition to governance. The structure of HL7 will be different from NCI or DOD. There is no single agreement - there is a series of agreements. Maybe HL7 needs lower level agreement.
   6. John Koisch: It is all doable - you have to pick your poison. HL7 builds according to ECCF, then say this is optimal, sub-optimal, doable under certain criteria. Ultimately the framework is very powerful - and should facilitate others work.
   7. Charlie Mead: To answer Andy, you look at the lifecycle, and look at what you can do to minimize risk.
   8. Steven Hufnagel: Put the products in the framework, then do a RASCI.
   9. Ron Parker:
      1. The enterprise we are developing a governance model for develops standards.
      2. The framework conceptualizes the production of the items from our mission and mandate.
      3. The consumers look at SAIF, see it as SOA based, and want to be able to get the information from HL7 on how to do it. To produce the specs, how do we organize and govern the business?
   10. John Koisch: Top-down architecture is fine. People are coming at it from useability. Governing the BF - it deals with correspondence with the other artifacts. SAIF provides an entry point.
   11. Ron Parker: SAIF provides the ability to produce products that fit BF. We are not doing the same thing that NCI/DOD are doing.
   12. John Koisch: we should re-draw the picture from HL7 Aspect. Bottom layer is domain committees.
   13. Ron Parker: Working Interperability problems are not owned by one domain. SAIF allows us to say when producing stuff, specs are coherent, rim-based, vocab enabled.
       1. How do we create GF in coherent way, and supported by WG's.
       2. We may have to look at how we group the organization.
   14. John Koisch: Binding between problem and solution. NCI does not facilitate arcitectural governance early. There is a problem with drive-by interoperability. There needs to be decomposition provided.
   15. Ron Parker: we develop a way to do things, and the reasons. NCI/DOD uses good concepts.
   16. Charlie Mead: We discussed HL7 having two architecture problems
       1. Internal architecture of HL7
       2. HL7 specifications that live in others architectures
          1. Has another way to think of governance
          2. We can control by our governance.
   17. Ron Parker:We need to influence WG's using the model correctly.
   18. John Koisch: The ECCF shows you how to categorize.
       1. People are trying to take it end-to-end, but that is not the point.
       2. The point is a solution specification is to specify the encapsulation barriers.
       3. The spec stack does not solve all of the problems.
   19. Grahame Grieve
       1. In our various cycles we had variance on what architecture should be.
       2. Success of architecture is not easily determined.
       3. There is not measure for HL7 Architecture
   20. John Koisch: Successful measure happens in other places.
   21. Ron Parker: Precepts include allow HL7 to function: There will be new working groups, and gives us more flexibility, coalesce them.
   22. Grahame Grieve: Success is a new WG
   23. Ron Parker: The organization can still proceed.
   24. Ron Parker:
       1. What are the tools/management
       2. Who is developing
       3. Rasci would be helpful in v2 & v3
   25. Jane Curry: We have not identified success criteria.
   26. Ron Parker: What does success look like?
   27. Charlie Mead: We bolted governance to risk reduction across the lifecycle.
   28. John Koisch:
       1. We produced specification to our idea, but it was not easily consumed.
       2. Process for referal generation: Presented slide that depicted boundaries between encapsulations. The is only one intersection between trading partners. Each encapsulation de-composes into contracts at all of the required levels.
       3. Identy management is inherent in the design. We built all of the interfaces to SAIF compliance.
   29. Ron Parker: This exposes the areas that need to be understood.
   30. John Koisch: DOmains have redundancy.
   31. Jane Curry: Required because of non-trusted interoperability.
   32. Ron Parker: Domain experts have to agree on how to persist the data.
   33. John Koisch: You may not want to expose as a service, instead a SQL call, which is Ok if it meets the requirements.
   34. Ron Parker: We need to provide governance that answers the question who is the expert, what is the data requirements?
   35. Charlie Mead: Johns chart by definition depends on governance across all of the entrprise.
   36. John Koisch: Patient Reg is non-cross funtional.
   37. Charlie Mead: TSC has dodged that this will crash head-on to the current organization.
   38. Ron Parker:
       1. We are talking about a GF wherein the organization can take on projects.
       2. Managing Risk: How can we enform managing the artifacts?
   39. Jane Curry:
       1. One of the key concepts of V3 was that the domain expert knowledge got captured, then constrained.
       2. In terms of vocab, domain model fragments, static-model-isch things that are aggregated are drawn from an inventory of assets.
   40. Ron Parker: Sub-artifacts
   41. Jane Curry: Artifacts have change management and are versioned.
   42. Ron Parker: How do we organize ourselves to accomplish the job? Competancies need to be granularized. e.g. OO or CDA are not going to own all of it.
   43. Jane Curry:
       1. We have people who produce reference artifacts.
       2. We have people who produced the technology artifacts
       3. We have domain people responsible for the requirements.
       4. Each of the WG's tries to manage all of the things in their use case.
   44. Steven Hufnagel: We need a set of use-cases.
   45. Ron Parker: It seems to me that a use case would be a requirement being asserted that registration is insufficient.
   46. Steven Hufnagel: It is a use case of how to form a project scope for a project scope.
   47. Jane Curry: Enable someone to use our artifacts as being coherent from both our point of view as well as the consumers's.
   48. John Koisch: HL7 organizes into steering divisions, which are above the problem space.
   49. Ron Parker: There are some basic organizing concepts.

Q2 Part 2

1. Andy Bond: We need to determine what we are governing against?
2. Ron Parker: We dont need an enterprise architecture for the WG, but use SAIF to discover the architecture.
3. Charlie Mead: John told the Board we would produce an architecture. We said no, we need an interoperability framework. We need to present them with the two architectures, RISK, etc, and tell them what we need to operationalize. Lets talk about operationalization. TSC does not have the framework.
4. Andy Bond: We have SAIF, but we need to leverage what we have.
5. Ron Parker: SAIF provides the context, but not a solution. We need to operationalize, then discover how to ballot.