DoD EA COI Data Exchange Standard I and SBSI do not have a vested interest in PES. Indeed, we developed it reluctantly. Our argument against XMI is based on our perception of the needs of the entire DoD architecture community – vendors, analysts, etc. and our concern about the negative impacts of over extending the XMI “standard”. DoDAF Team 28 March 2011
Topics Bird’s eye view of DoD EA Data Exchange PES and DoDAF XMI DoDAF CM and UPDM Conclusion Recommendations
Why would a DoD architect exchange data? Across For reuse Might be convenient but often more trouble than it’s worth Downward To bootstrap a refinement Might be convenient Upward To integrate and analyze Otherwise it’s “wet” information fusion and ad-hoc analysis Most DoD architecture “exchanges” today are pdf, ppt, doc, xls, html, SA encyclopedias, … -- not optimal, most participants would prefer otherwise
DoD EA Data Exchange Standard Specialty: JACAE Specialty: XMI Federal: OMB Cross-Agency Specialty: TBD Specialty: CUDEAM Allied NATO Coalition Specialty: BPMN Specialty: others Specialty: SA Encyclopedia Specialty: Archimate Goal: minimize translation Current givens: DoDAF 2 is founded on DM2 DM2 and CUDEAM are founded on IDEAS Consequent: the DoD EA data exchange standard should be as semantically close to DM2 and IDEAS as possible
COTS Architecture Tools Vendor’s Day 14 March 2011 + Follow-up (page 1 of 2)
COTS Architecture Tools Vendor’s Day 14 March 2011 + Follow-up (page 2 of 2) We have just started looking into relevant Govt-developed tools and ADS We have not figured out a way to know which federated architectures are currently or planning to use which tools.
Lay of DoDAF Land Model specifications (AV-1 SV-10c) DM2 Intent was to specify using data dictionary terms Unfortunately too much legacy committee language was preserved Estimate 75% of the terminology is undefined and inconsistent This ambiguous specification, accumulated since 1994, has led to the evolution of an EA community disconnected from DoD’s six core processes DoDAF is intended to support DM2 Conceptual Data Model – very simple Logical Data Model Because of IDEAS there are only ~250 total data elements compared to the less-expressive CADM that had ~16,000! But IDEAS is hard to learn – it’s mathematical Physical Exchange Specification is Automatically generated from the LDM Slightly-dumbed-down LDM in XML so if you know the LDM, PES is simple PES tags and definitions are identical to DM2 LDM No new structures are introduced other than XML-isms MITRE on the UPDM Team has been opposed to the idea of PES since it’s inception 3 years ago The 52 DoDAF models and the DM2 are related via a matrix* * 52 DoDAF models X 250 DM2 data elements, referred to as the “monster matrix” because it has ~ 13,000 decision cells
PES Structure Made for future interoperability with IDEAS-based data Packaging, e.g., overall classification marking Extra goodies for Dublin Core (optional) Actual architecture data – tag names and definitions are exactly from DM2 LDM Where you say what views the data corresponds to One PES file can have multiple views A single piece of data can be in multiple views A recipient of the XML file should validated it against PES XSD which automatically encodes the “monster matrix”. XML stuff -- unimportant
XMI Very complicated – has taken years for a very limited set of like-functioned CASE tools to exchange UML meta model carries a lot of early Object Oriented programming baggage and was designed by committee. Consequently, it is badly suited to EA. This was the biggest problem in using UML in M3. Specific to UML tools, ~ 10% of the tools identified by the A&I team We have not yet found any DoDAF XMI files in DARS or NARS Sweden estimated UML tools used for less than 1% of their EA’s Most architecture tools we have identified are not UML There will probably never be a business case for them to implement XMI, esp. given how costly it has been for the tools in the CASE market There are other modeling language standards, e.g., BPMN, Archimate In many cases it is not even possible for the tools to conform to XMI because they are not software engineering tools I.e., OMG standards do not apply to 90% of architecture tools An IEEE, INCOSE, or ISO standard would be broader
Compare: UPDM Search and Rescue OV-2 Example NOTE: In this example, UPDM Team disagrees with early DM2 LDM decision that input and output happens only a result of an activity. This was done to fix to long-standing CADM problem. UPDM does not have a solution to this problem and so it is possible to create inconsistent data as in CADM.
PES vs XMI PES 217 lines Almost all are direct DM2 terms so anyone familiar with DoDAF and DM2 can read XMI 1,272 lines* Almost all are UML terms so only someone familiar with UML could read * Much bigger even with the dangerous shortcut noted in the prior slide See handouts. There are many samples on the DoDAF Journal site. GFE software on the WG site can generate PES easily.
UPDM Team Participation in DoDAF-DM2 Configuration Management Not where the issues are UPDM Team has been a DoDAF-DM2 Working Group member for over two years Approximately 75 UPDM Team Change Requests have been submitted, most of which were incorporated in v2.01 and v2.02 Only three on PES Two the WG deferred to v2.04 One new one is on “monster matrix” (entered by Greg Schaefer) UPDM Team knows our Greg Schafer has been their resource for PES and IDEAS Foundation questions and aid for over two years Small only because WG is reluctant to turn attention to the DoDAF model specifications.
DoDAF-DM2 CM Process Involves Peer and Component Reviews DoDAF Development Team does not control DoDAF or DM2 400 member WG does peer review and work FAC/ASRG oversees and coordinates formal Component review and approval of new versions
WG Business Rules Have Been Essential for Broad Community Consensus Some people don’t like the rigor. It is hard but the alternatives are worse.
DoDAF-DM2 Conformance (proposed to FAC as change for v2.03) Levels Conceptually conformant Uses DoDAF terms and aliases (from DM2 CDM) to categorize its concepts DoDAF views (AV-1 thru DIV-3) have correct information according to “monster matrix”). Logically conformant Level 1 + adheres to terms and relationships from DM2 LDM and aliases Physically conformant Level 2 + expressed as DoDAF – DM2 PES that can be consumed by others Semantically conformant Level 3 + IDEAS semantics are correct Confirmation method Inspection Test of XML files against standard schema validator TBD, but would mostly likely be a test of the OWL/RDFS files e.g., USAF RA was deemed Level 2 Vendors may wince at levels 3 and 4 because they are testable Gives vendors and architects flexibility – FFP!
Conclusion PES is not separate from or more complicated than DM2 PES is just a way to exchange DM2 data via XML DM2 is the only solid part of DoDAF 2.0 The DoDAF model descriptions have major quality problems which will be a drag on any framework consolidation efforts -- Federal, Coalition, NATO, … Using DM2 is the only hope of fixing these problems. DM2 CDM is being used by Intelligence Community JAWG DM2 OWL is being used by Business Mission Area for its Semantic Web DM2 PES has been developed and exchanged in a few pilots and examples, e.g., JACAE, CADIE, MCAE, NARS There is currently no broad-based commercial standard for architecture data exchange XMI is not it Most (90%) tools do not and probably cannot support Most Govt tools and ADS would not be able to import or export XMI If a single commercial standard is desired, a broader base is needed, e.g., IEEE, INCOSE, ISO, IDEAS, and/or NATO
Recommendations Continue work with IDEAS Group and NATO to develop CUDEAM with goal to move into a broad-based standards body, e.g., ISO, IEEE At the DoD level, use the 400+ DoDAF WG members for peer review – includes all vendors as well as UPDM team Moving to international, remember AP-233 and Eurostep IDEAS methodology and Business Rules will help Support all vendors Do not mandate a small segment standard that would disqualify 90% of the market Define a Level 3.5 conformance: PES generation only Remind vendors Level 2 may be sufficient for some of their clients – FFP! Encourage use of XMI, BPMN, Archimate, etc. for specialized exchange between appropriate tools DoDAF WG peer review the UPDM PES white paper when it becomes available