soa-forum
[Top] [All Lists]

RE: [soa-forum] The purpose of a SOA demo

To: "Cory Casanave" <cbc@xxxxxxxxxxxxxxxxxxxxxxx>, "Service-Oriented Architecture CoP" <soa-forum@xxxxxxxxxxxxxx>
Cc: "John F. Sowa" <sowa@xxxxxxxxxxx>
From: "Paul Prueitt (ontologystream)" <psp@xxxxxxxxxxxxxxxxxx>
Date: Mon, 27 Mar 2006 09:40:12 -0800
Message-id: <001701c651c5$83c7e2c0$4064a8c0@YOUR8FE0F439A7>

Cory,

 

I will spend the next 24 hours reading and thinking carefully about the materials I find at     

 

www.osera.modeldriven.org

 

I have been aware of your effort but not the specifics.

 

Cyrille’s work on extending BCM focuses on some common issues… such as “which technology should be used”

 

        “usage of other technologies like ESB (with WS-RM & WS-Notification), ebSOA, SEE’

 

 

In a very broad use case, one might start out with several groups (such as ourselves) being separated by our use of language and individual conceptualizations.  How can collaboration/interoperability be developed that STARTS with a deep analysis of the full set of issues (the interaction space).  What services are needed to facilitate that collaboration/interoperability?

 

Wow!!

 

 

Joe and Farrukh , for example obviously have great respect of the other’s work,  but somehow there is not an agreement on something that is a core issue.  So is there a service that is needed here?  What would this service be called?

 

Reconciliation of terminology issues is one way to tease out these differences between individual conceptualizations and to then allow a synthesis based on a deeper (commonly felt) _expression_ of purpose.

 

To a very great extent, Cyrille’s and mine approaches are becoming coherent (singular) through appreciation and through a focus on larger purposes.

 

Several individual “conceptualization” are being synthesized into a “BCM business” layer. 

 

Having said that, our industry fails to move from high level conceptualizations to implementation.   The substance of all of the individual conceptualization, and the substance of the merged conceptualization is not “actionable” as yet. 

 

This is what BCM is all about.

 

Along the way, we should be able to find precisely where the OsEra makes a contribution and where that may be some evolution pathway interesting to OMG (Object Management Group).

 

Many in this forum have come to understand  my critic of FEA (federal enterprise architecture) as being too focused on what IT vendors are comfortable with and not focused on taking several steps towards the BCM notion of choice points (contracts between humans (at the conceptual layer) that often radically simplify what the technology has to do).

 

This simplification (if demonstrated) is where the huge value proposition is at.  !!!!!!

 

An example, just to be specific, is in the SOA notions of “web service discovery” and “web service orchestration”. 

 

What about “service discovery” and “process orchestration” without the “web” adjective? 

 

Humans do ‘service discovery” and “process orchestration” sometimes in a masterful way, and sometimes very poorly.  But in the general case, no computer program comes close to the synthesis that is required to create / discover what needs to be done.

 

Perhaps this is why government response to Katrina was as it was.

 

 

 

 

 

 

 


From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of Cory Casanave
Sent: Monday, March 27, 2006 8:36 AM
To: 'Service-Oriented Architecture CoP'
Subject: RE: [soa-forum] The purpose of a SOA demo

 

Paul,

One of the capabilities we have been working on that we would like to showcase as part of our contribution to the demo is the ability to render the same architectures in multiple forms, including OWL (Description logic) – as well as integrate these multiple forms of expressing architecture, bi-directionally.  Starting from a business architecture (In our case using MDA technologies) we could then show how it can map to the technology stack as well as other ways to see produce the same information as an ontology.  Much of this is based on the “OsEra” work; osera.modeldriven.org.  The same approach could also be used to map to other kinds of representations.

-Cory

 


From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of Paul Prueitt (ontologystream)
Sent: Monday, March 27, 2006 9:44 AM
To: 'Service-Oriented Architecture CoP'
Subject: [soa-forum] The purpose of a SOA demo

 

Joe,

 

Excellent article at

 

http://www.ebxmlforum.net/articles/ebFor_20030824.html

 

It would seem that you, Cory, Andrew, Farrukh and Rex have positions (regarding registry and repository) that are very close and saying (for various good reasons) that ebXML should interoperate with UDDI because it can and because of the previous market adoption of UDDI.   You are also saying (I conjecture) that there is no other “registry/data definition” standard that needs to be considered? 

 

The exception is ontology representations which I believe there is an agreement that ebXML, but not UDDI, allows an interface to any

 

1)   description logic

2)   object oriented model like UML or (say) SOA-IM (from Semantion)

3)   RDBMS type data definitions (and SQL) 

4)   N-ary ontology as still theoretical work, or in same case (NdCore from www.apptechsys.com , other “semantic extraction ontology internal representations)

 

 

Is this fair and proper to say?   Do all concur?

 

Note that I am explicitly placing (part of) “data definition” into the “repository” technology.  But an OWL ontology would not be in the repository, if I have a proper understanding of ebXML repository concept.   I assume that there is no misunderstanding. 

 

As we will see, the core issue that is not talked about is how are data definitions created. 

 

Is there re-use, granularity, agility, composition? 

 

Where is the origin of design? 

 

Do IT community design the web-services or do a (perhaps) more full native stakeholder (non-IT business person, government policy maker, natural scientist) use a methodology to define natural needs and means to respond to a request for assistance?

 

Is the demo you all are contemplating IT-centric, and/or does it really matter?

 

One may take the position that how data definitions (in the broadest scope) come about could be examined more carefully from a knowledge management type position (ie not a pure IT perspective … having more of a business centric methodology perspective – as in OASIS BCM 2004).

 

This leads to the notion of a choice point, when the user is given an informed opportunity to make a decision based on (to a large extent) non-IT issues.  Often this goes to the nature of purpose. 

 

Yes, the purpose of a web service can be defined, and in fact one can work hard to show some degree of agility, composition, re-use and granularity.  But is there a common dimension to natural social interaction that is not present in the proposed demo?   

 

Is this important, if there is a difference?

 

 

This is the service part of the SOA, where “service” is not sameAs “web-service”, but is in fact a naturally occurring interaction type between human and human communities.  The choice points can be / are where the design of web services occur.  I would suggest that ebXML does not fully realize this possibility.  Perhaps this is where a new discussion, and OASIS specification needs to be focused.  If so, then perhaps the mature part of KM should be referenced?

 

 

We imagine that there might be some additional discussion, and if so perhaps just the topics might be of value; but not a detailed discussion?

 

One is about the very definition of “registry” and “repository”….  Most know that there are non-standard variations to these technologies – some of them quite radically different.  But this topic is off the table at least in the context of Federal US CIO Council discussions.

 

Questions

 

Does ebXML has native ability to interact with description logic based ontology?  Are there any reservations about this?

 

Is there a complete ability to call from ebXML to a stored procedure or a functional program, (something with inputs and outputs)?

 

In the XML accelerators, (in reference to the SUN READ technology, do you (Farrukh) see a deep and seamless address-ment of the band weigh issues (streaming compressed XML has high costs).  Also READ seems to reduce the need for memory allocation by making a data compression type dictionary so that the “data” and the “data structure” are written in isomorphic fashion but with less bits and bytes. 

 

Would you briefly describe READ (representation something something) unfortunately Google does not help here..  I could not re-find the paper you referenced.

 

I am just trying to verify where we are in this discussion.

 

Comments?

 

 

 


From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of Chiusano Joseph
Sent: Sunday, March 26, 2006 7:15 AM
To: Service-Oriented Architecture CoP
Subject: RE: [soa-forum] SOA Demo Straw men (Plural)

 

Speaking neutrally, and not intending to spark a debate:):

 

In case it may help our understanding of ebXML Registry and UDDI, here an article that I wrote 3 years ago (April 2003) called "UDDI and ebXML Registry: A Co-Existence Paradigm" that speaks to both registry types, and a vision for co-existence between them. 

 

At the time of the article, the market was very fragmented in respect to these two registry types - since then, we have seen great changes. For example, there are several vendors who now offer both support for both ebXML Registry and UDDI in their products. The article explains how ebXML Registry and UDDI overlap in some capability areas (service registry), and the additional features that ebXML Registry provides (of most interest for our purposes here) - as well as some market predictions (of least interest for our purposes here).

 

Here is the article:

 

 

For further interest: Here is also a followup article that described how (at the time of its writing, Sept 2003) both registry standards could seamlessly interact, as well as a broader vision for registry interoperability:

 

http://www.ebxmlforum.net/articles/ebFor_20030824.html

 

Please followup with me offline if there are any questions on these that are not pertinent to our current discussion here.

 

Thanks,

Joe

 

Joseph Chiusano

Associate

Booz Allen Hamilton

 

700 13th St. NW, Suite 1100

Washington, DC 20005

O: 202-508-6514 

C: 202-251-0731

Visit us online@ http://www.boozallen.com

 

 


From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of Cory Casanave
Sent: Sunday, March 26, 2006 9:39 AM
To: 'Service-Oriented Architecture CoP'
Subject: RE: [soa-forum] SOA Demo Straw men (Plural)

Farrukh,

I don’t have a strong requirement for UDDI, we can see what others may think.  I can imagine 3 approaches for the community;

* ebXML

* UDDI

* A community-specific service, perhaps wrapping one of the above.

 

I haven’t looked at ebXML deeply in quite some time (I was part of the initial release, mostly on the BPSS).  I very much like the model but couldn’t get much vendor support and thus the customer interest waned.  Does use of the registry imply that ebXML must be used for the other roles?  How is ebXML integration with non-ebxml services?

 

-Cory

 


From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of Farrukh Najmi
Sent: Sunday, March 26, 2006 8:30 AM
To: Service-Oriented Architecture CoP
Subject: Re: [soa-forum] SOA Demo Straw men (Plural)

 

Cory Casanave wrote:

Farrukh,
Its on the Wiki at;
http://colab.cim3.net/file/work/SOACoP/SOA%20Community%20of%20Practice%20Dem
o.doc
  


<excerpt from doc>

Scenarios

Registration of a new actor in the community

A simple UDDI based service at a well known community portal


</excerpt from doc>


Is there a reason to be prescribing UDDI above? I would like to participate with my open source freebXML Registry [1] , [2] which implements the OASIS ebXML Registry standard [4] (not UDDI). The ebXML Registry standard is also an ISO standard. It has a super set of functionality than UDDI  [8].

There is a growing realization that a registry without an integrated repository is inadequate for SOA [9] and that ebXML Registry is better suited for complex SOA dpeloyments [10]. The basic premise is that "you can't govern what you don't control". If artifacts are stored external to an integrated repository then they cannot be effectively governed. I have made this very point in [11].

Please let me know if above is a requirement or not. It does not seem to be a requirement with IRS, DISA / DoD etc.

Thanks in advance for any clarifications.

[1] freebXML Registry
<http://ebxmlrr.sourceforge.net>

[2] Reference Deployments of freebXML Registry
<http://ebxmlrr.sourceforge.net/aboutFAQ/About_freebXML_Registry.html#Deployments>

[4] ebXML Registry 3.0 Specifications Bundle (OASIS Approved Standard)
<http://www.oasis-open.org/committees/document.php?document_id=12049> (the specifications)
<http://xml.coverpages.org/ni2005-02-14-a.html >(Cover pages article detailing new functionality in version 3.0)

[8] LDAP, UDDI and ebXML Registry feature comparison matrix
<http://ebxmlrr.sourceforge.net/tmp/Registry_Capability_Matrix.html>

[9] UDDI Inadequate for SOA:
<http://www.it-analysis.com/article.php?articleid=12909&SESSID=3481b206b49825751535e90741ec4cb2>

[10] ebXML registry suited for SOA complexities, say proponents:
<http://searchwebservices.techtarget.com/originalContent/0,289142,sid26_gci1122135,00.html?track=NL-110&ad=527456>

[11] SOA Rgeistry White Paper:
<http://www.sun.com/products/soa/registry/soa_registry_wp.pdf>

-- 
Regards,
Farrukh
 
 _________________________________________________________________
Subscribe/Unsubscribe/Config: http://colab.cim3.net/mailman/listinfo/soa-forum/
Shared Files: http://colab.cim3.net/file/work/soa/
Community Portal: http://colab.cim3.net/
Community Wiki: http://colab.cim3.net/cgi-bin/wiki.pl?AnnouncementofSOACoP    (01)
<Prev in Thread] Current Thread [Next in Thread>