[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [soa-rm] Service-Orientation, SOA, RM vs. RA, etc.: Suggestion To Bring Us Closer Together
I think that we are not going to end up with a simple conformance guide for SOA: if you pass *our* test then you have an SOA. Although we *could* declare that XML/WSDL/BPEL/WS-*/* is what defines SOA, a lot of people would be unsatisfied by such a definition because it is inherently ephemeral. I personally think that it is important to be able to capture the fact that SOAs tend to be about having multiple interacting services. Preferably using transparent service combinations. However, in modeling a conversation, there are two approaches: focus on the people talking, or focusing on what they are saying. The former analysis tends to result in groups of people, the latter on what makes an utterance comprehensible. The latter kind of analysis absolutely assumes that more than one person is talking; it just does not need to belabor the point. Frank On May 20, 2005, at 8:55 AM, Matthew MacKenzie wrote: > What if a core value of your architecture is that all consumers > somehow are seeded with knowledge of everyone's FTP server..e.g. > (Bonjour/zeroconf/mdns)? > > -matt > On 20-May-05, at 11:47 AM, Michael Stiefel wrote: > > >> Or to be concrete about it. If I put some WSDL for my Web service >> on an FTP site, I have a SOA according to our current definition. >> I too find that intuitively difficult. >> >> Michael >> >> At 09:51 AM 5/20/2005, Christopher Bashioum wrote: >> >> >>> Matt, >>> >>> Your response below was excellent. I especially liked the >>> following: >>> >>> <quote>SO = Philosophy, and hopefully doctrine >>> eventually. SOA, on the other hand, is practice and adherence to >>> our >>> doctrine. </quote> >>> >>> This being the case, then, there is value in creating something >>> that can be >>> held up to an existing architecture to determine if that >>> architecture >>> follows the SO doctrine. >>> >>> For example, if I expose a bunch of application functionality to >>> the world >>> via web services, does that constitute an SOA? The answer based >>> on our RM >>> so far would be 'No', because there is no service description >>> captured in >>> metadata that enables discoverability. However, if I document >>> the web >>> service in an interface control document and store that document >>> in a public >>> folder somewhere, do I now have an SOA? Based on our RM so far, >>> the answer >>> would be 'Yes' ( which bothers me, as intuitively I don't believe >>> that is an >>> SOA) >>> >>> -----Original Message----- >>> From: Matthew MacKenzie [mailto:mattm@adobe.com] >>> Sent: Friday, May 20, 2005 7:33 AM >>> To: SOA-RM >>> Subject: Re: [soa-rm] Service-Orientation, SOA, RM vs. RA, etc.: >>> Suggestion >>> To Bring Us Closer Together >>> >>> Joe, >>> >>> In reading this thread, I noticed your question re: SO vs. SOA. I >>> think this is why the question: >>> >>> "Service Oriented Architecture Reference Model" vs. "Reference Model >>> for Service Oriented Architectures" tweaked in my head a few weeks >>> ago. I find myself typing SO more than SOA lately, and >>> Hamid...despite the fact that I am not seeing things in his vision, >>> has triggered something in my brain with regards to OO. >>> >>> Contrasting SO to OO is probably a useful approach. I view our work >>> here as being largely theoretical, which really does put us in line >>> with a concept such as OO, which really does not touch language and >>> implementation issues. >>> >>> I really would like to throw out consideration of "RA" completely. >>> If we do define any architecture, if an RM can indeed be >>> construed as >>> an Architecture, it would be a transcendental architecture -- almost >>> spiritual in nature. The most interesting thing I have read all >>> week >>> was a post by Frank on how those of us sitting close to the >>> theoretical realm of computer science are basically philosophers >>> more >>> than anything else. SO = Philosophy, and hopefully doctrine >>> eventually. SOA, on the other hand, is practice and adherence to >>> our >>> doctrine. >>> >>> "And are not those who are verily and indeed wanting in the >>> knowledge >>> of the true being of each thing, and who have in their souls no >>> clear >>> pattern, and are unable as with a painter's eye to look at the >>> absolute truth and to that original to repair, and having perfect >>> vision of the other world to order the laws about beauty, goodness, >>> justice in this, if not already ordered, and to guard and preserve >>> the order of them--are not such persons, I ask, simply blind?" >>> --Plato, from Republic >>> >>> Now, the point I am making is not that focusing on architecture is >>> stupid. My point is that a higher order of understanding is >>> required >>> to form a basis for future work. >>> >>> Isn't it glorious to be a philosopher-king? >>> >>> -Matt (who is amazed that his liberal arts education is useful in >>> his >>> chosen field) >>> >>> >>> On 20-May-05, at 6:44 AM, Chiusano Joseph wrote: >>> >>> > <Quote> >>> > Does that work for you? >>> > </Quote> >>> > >>> > Not at all. The issue is not about the charter (at least not >>> > primarily). I would simply like to see us address the questions >>> > that I proposed in the "pulse check" to get a sense of how the TC >>> > feels as a whole about these fundamental issues. A charter can say >>> > "we are to develop X" and "here is what X is", but if - having >>> said >>> > that - when the work begins, it becomes clear that there are still >>> > places within the charter where there are room for interpretation, >>> > and the interpretation is not unified, I believe it is >>> justified to >>> > have clarification. >>> > >>> > Then if we see that the majority of the TC members are in the "I'm >>> > fine - please proceed" category, there is no issue. If there is, >>> > then we should go down the path of re-examining the charter. But >>> > that may not even be necessary. >>> > >>> > Thanks, >>> > Joe >>> > >>> > From: Duane Nickull [mailto:dnickull@adobe.com] >>> > Sent: Thu 5/19/2005 11:23 PM >>> > Cc: soa-rm@lists.oasis-open.org >>> > Subject: Re: [soa-rm] Service-Orientation, SOA, RM vs. RA, etc.: >>> > Suggestion To Bring Us Closer Together >>> > >>> > Joseph: >>> > >>> > I have been aware of only a few who are wanting to re-examine our >>> > charter. Nevertheless, we are democratic. We will put this up >>> for a >>> > vote. If more than one third of the members feel this is worth >>> taking >>> > time on, we will discuss it. The one third represents the fact >>> that >>> > some may not actually vote. If less than one third select to >>> discuss >>> > it, then can we please accept the charter? >>> > >>> > We will set it up tomorrow and leave it open for one week. That >>> > leaves >>> > plenty of time if it passes to distribute the questions then >>> > compile the >>> > results. >>> > >>> > The rationale is that while a few may still wish to examine it, my >>> > perception is that the vast majority do accept the charter and >>> want to >>> > work on a reference model first, then RA. >>> > >>> > Does that work for you? >>> > >>> > Duane >>> > >>> > >>> > Chiusano Joseph wrote: >>> > >>> > >So you don't see any problems regarding any of the below? >>> You're not >>> > >aware of anyone expressing concern on our list regarding what it >>> > is we >>> > >are defining, the scope of it? Hmmmm.....maybe I've been >>> operating >>> > in a >>> > >different TC.;) >>> > > >>> > >Oh - I am also not referring to what is in our charter. "It is >>> in our >>> > >charter" is not, IMO, an effective way to address the concerns >>> that >>> > >people have been repeatedly expressing). One can put things in a >>> > >charter, then start work, and be unclear as to whether or not >>> they >>> > have >>> > >strayed from the charter. >>> > > >>> > >This is a simple request from a TC member to clarify what they >>> are >>> > >perceiving is a major disconnect within the TC on several issues, >>> > and it >>> > >seems that the answer from the Chair on that is "I don't see any >>> > issue", >>> > >when I believe it should be "Let's address these concerns". >>> > > >>> > >Thanks, >>> > >Joe >>> > > >>> > >Joseph Chiusano >>> > >Booz Allen Hamilton >>> > >Visit us online@ http://www.boozallen.com >>> > > >>> > > >>> > > >>> > > >>> > >>-----Original Message----- >>> > >>From: Duane Nickull [mailto:dnickull@adobe.com] >>> > >>Sent: Thursday, May 19, 2005 11:03 PM >>> > >>To: Chiusano Joseph >>> > >>Cc: soa-rm@lists.oasis-open.org >>> > >>Subject: Re: [soa-rm] Service-Orientation, SOA, RM vs. RA, >>> > >>etc.: Suggestion To Bring Us Closer Together >>> > >> >>> > >>Comments inline: >>> > >> >>> > >>Chiusano Joseph wrote: >>> > >> >>> > >> >>> > >> >>> > >>>Duane, >>> > >>> >>> > >>>I would like to make a suggestion to help clear up the current >>> > >>>division in our TC on some basic issues, which I believe is >>> truly >>> > >>>inhibiting our ability to move forward in a unified way - >>> and will >>> > >>>continue to do so unless we address it at this time. >>> > >>> >>> > >>>The most prominent division that I have perceived over the >>> > >>> >>> > >>> >>> > >>course of >>> > >> >>> > >> >>> > >>>several weeks is: "If we are defining a reference model, >>> what is it >>> > >>>for? Is it for a single service? (call this >>> > >>> >>> > >>> >>> > >>"service-orientation") or >>> > >> >>> > >> >>> > >>>SOA?" IOW, "Is it SO-RM, or SOA-RM?" >>> > >>> >>> > >>> >>> > >>I think that there is no disagreement of what a RM is or what >>> > >>we are calling the TC. That has been specified in the >>> > >>charter from day 1 in very clear language. We did have a >>> > >>brief conversation about the name but it was my observation >>> > >>that only 1 or 2 were even willing to change it. The rest of >>> > >>the 91 members seem to be in agreement. Likewise - who is >>> > >>still confused as to the purpose of a reference model? >>> > >> >>> > >> >>> > >> >>> > >>> >>> > >>>The second most prominent division that I have perceived >>> over the >>> > >>>course of several weeks is: "Where is the line drawn between >>> RM and >>> > >>>RA?". Last week I began a thread[1] on this question, and I >>> > >>> >>> > >>> >>> > >>thank all >>> > >> >>> > >> >>> > >>>who contributed (Matt, Duane, Ken, Rex, Francis, any others >>> > >>> >>> > >>> >>> > >>I missed). >>> > >> >>> > >> >>> > >>>However, I think we really need to drill down into this >>> > >>> >>> > >>> >>> > >>question more >>> > >> >>> > >> >>> > >>>and have a crystal clear answer before we go any farther, >>> > >>> >>> > >>> >>> > >>else run the >>> > >> >>> > >> >>> > >>>risk of creating an RM that cannot easily "bridge to" an RA. >>> > >>> >>> > >>> >>> > >>This is something that is less clear but I feel we are on >>> > >>track with our current activities. Matt's email clarified it >>> > >>very well IMO. We now have a collective responsibility to >>> > >>ensure our RM is usable, unique etc. We must be vigilant in >>> > >>that regard. >>> > >> >>> > >> >>> > >> >>> > >>> >>> > >>>So, I would like to propose a solution: >>> > >>> >>> > >>>I would like to propose that we take an informal poll (not a >>> formal >>> > >>>vote) across the TC as a "pulse check" that will enable us >>> to come >>> > >>>closer together on these vital issues. The poll would be >>> > >>> >>> > >>> >>> > >>comprised of >>> > >> >>> > >> >>> > >>>the following questions (folks would simply put an "*" to >>> > >>> >>> > >>> >>> > >>the left of >>> > >> >>> > >> >>> > >>>the letter of their response): >>> > >>> >>> > >>><Questions> >>> > >>>(1) Do you believe that the RM in our current draft is: >>> > >>> >>> > >>>A. A service-orientation reference model B. A SOA reference >>> > >>> >>> > >>> >>> > >>model C. >>> > >> >>> > >> >>> > >>>Other >>> > >>> >>> > >>> >>> > >>Joseph - I am sorry but this is in our charter. it is not up >>> > >>for negotiation. Everyone who joined this TC had the >>> > >>opportunity to read the charter. We allowed discussion on it >>> > >>once or twice and my recollection is that there is clear >>> > >>consensus on both the name and purpose of the TC. >>> > >> >>> > >>Reference Models are clearly scoped and defined. This TC >>> > >>should not impose to re-define what a reference model is. >>> > >>First - it will probably not fly with established software >>> > >>architects. Second - we already decided to adopt and use the >>> > >>industry standard definition (again - in the charter). >>> > >> >>> > >>We have much more important work to contemplate. I would >>> > >>like to harness the collective experience and energy of this >>> > >>TC to get the core model nailed down. >>> > >> >>> > >>As we progress, we will have the opportunity to examine and >>> > >>tune the RM to be useful. >>> > >> >>> > >>Duane >>> > >> >>> > >> >>> > >> >>> > >>> >>> > >>> >>> > >>> >>> >> >> >> >> > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]