Hi Cory, (01)
Ok, I take your points, but I'm wondering how much the community aspect
of the SOA is influenced by the technology capabilities and programming
paradigms supported by an MDA approach. I've been going back through
some of the original mails around this, and been doing some thinking. (02)
Maybe we can try a collaborative approach to scenario building? I know
you didn't like my last demo proposal very much, so I'm willing to work
with you here on the buyer/broker/seller[manufacturer] critter. (03)
However, I think the best place to start here is with the following
requirements: (04)
1) The system should demonstrate interoperability between community
participants. Interoperability in this case is defined by whatever
automated systems communicating using specifications and protocols are
agreed by the community. Do we have a firm list? (05)
2) The system, to be realistic, must support user-initiated and
automatic purchase requests. In this way, we can provide an interactive
demo which allows for more real-world scenarios such as someone special
ordering something as well as regular or trigger-based, automated order
requests. (06)
3) Orders must be requested by supplying the following information: (07)
- the purchaser identity
- the number and quantity of the (single) item
- the amount the purchaser is willing to pay
- the timeframe in which the items must be delivered
- the timeframe in which the availability of the item must be confirmed (08)
4) If the order cannot be confirmed within the timeframe, the purchaser
must be given a list product descriptions and prices of either a)
alternative products which are "similar" or b) products which are
outside what the purchaser has bid. (09)
5) Purchasers must manually register with the broker, supplying normal
contact information. As part of this registration process, they will be
identity-proofed by the broker's human staff. If this is completed
satisfactorily, the purchaser is issued with an X.509 digital
certificate which must be used when making any purchases. It is assumed
that the purchaser has also established a line of credit with the broker
at this stage, as requested during the registration process. (010)
6) Sellers/manufacturers must register with the broker in a similar
manner to purchasers. As part of the registration process, the
seller/manufacturer will specify: (011)
- SLA for responding to requests for goods
- SLA for delivery of goods
- The interval at which the catalog will be updated
- The manner in which the seller/manufacturer is contacted for quotes
and final prices. (012)
If this is completed successfully, the seller/manufacturer is issued
with an X.509 certificate to authenticate it to the broker. (013)
7) All service brokers must register themselves at a well-known
location. (014)
8) When a broker receives an order, it will first compare the order
criteria with its internal catalog to find appropriate supplier(s), and
then confirm that the supplier has the item in stock and can supply it
to the purchaser within the specified criteria. (015)
9) If any problems occur with the order after it has been made, the
seller/manufacturer must notify the broker within timescale X. If so
notified, the broker must notify the purchaser. (016)
10) Purchasers can cancel orders within timescale X of placing it, or
prior to shipment. (017)
11) Purchasers can request a complete catalog of the available products
offered by the broker either interactively or in an automated fashion. (018)
12) Once created, purchasers can check the status of their order based
on a confirmation order number. (019)
Then we can talk about message formats, protocols and invocation
mechanisms. I could easily see this system being implemented using a
number of technologies, but I would also say that you could quite
happily do it with something like XML(Atom/1.0+custom vocabulary), XSLT,
HTTPS a standard web application and X.509, cert-based authentication
and a Web browser. (020)
This solution would meet the requirements, and it would be completely
interoperable, be browser-friendly and would also allow things like
standard, off the shelf syndication readers to be leveraged in extending
the services offered. These are the benefits of standardized formats
and invocation interfaces. However, it wouldn't use WS-* at all, which,
as I understand it, is a requirement. (021)
If the architecture reflected in the current document is intended to
reflect these business requirements, I would argue that it is too
closely aligned with fine grained messages that aren't self-describing
(this doesn't mean schema, this means stand-alone/context-free) and
which aren't directly aligned with the types of interactions supported
by humans or real-world entities. (022)
Today, you would have someone give purchasing requirements to a broker,
and then they would either say "yea", "nay" or "we can't do it, but
here's your options." I would see this as a business message, self
describing, and exchanged between the purchaser and the buyer. It
naturally aligns itself to the existing business model. (023)
I realize that this model is a bit looser than the one you propose, but
if we say that the numbered items represent the business requirements,
we can demonstrate they can be fulfilled in a service-oriented manner
using a variety of technologies and platforms. You still have the
community, you still have the choreography between participants, you
still have the data and you still have the constraints. However, you
also can then say things like "these are the minimum items you must
supply, but if you want to supply more (like photos), you can" to your
manufacturers. (024)
I would see this as a pretty interoperable community which also allows
collaboration and new applications to be built on an agreed foundation
of messages, addresses (URIs) and invocation mechanisms. If you want
reliable delivery of messages between participants, that's cool, we can
do that in a variety of ways. If you want to transform vocabularies
between broker/buyer or broker/manufacturer, then insert an intermediary
service. If you want management, then specify statistics messages to
receive, how and where. (025)
I also think that the scalability requirements only apply to the
registry of brokers and the brokers themselves. The buyers should
interact with the broker in one direction. The manufacturers and the
broker should have two-way communication, but if the manufacturer was
unavailable, that's ok. They loose the business. No big deal to the
community. If there is no alternative, there's no order, or the order
can be retried within the timeout to ensure that the manufacturers have
a chance to get back on-line. (026)
What does everyone think about the above as the revised business
requirements for the buyer/broker/seller scenario? If there's
agreement, then we can prioritize the requirements to focus on those
that are of most interest, then we can formalize the business
relationships/messages/protocols within a revised demo document (which
is technology neutral except for those technologies deemed necessary for
the demo). Finally, participants create solution documents describing
how they will meet those requirements with their tools/technologies. (027)
I don't think the on-the-fly or 1hr wrapping an existing service aspects
are really anything other than "dog and pony show" type things. They
aren't really going to help you much in a real SOA environment unless
you're really lucky, or there's some aspect to it that I'm not aware
of. It might be cool, but I'm not sure that's part of what we want to
show if we're talking about legitimate business value to stakeholders. (028)
As I said, I'm open to making this a collaborative effort. I have tried
to capture the intent of Cory's specification without the technology
aspects in this email (where possible). I am happy to propose that we
collect from here into an on-going specification somewhere. I'm tempted
to volunteer to do this, but then I will get even further behind with my
already maxed-out workload between now and the 22nd. (029)
I guess you can feel free to propose something else, or say I should
just give up and we'll go with whatever Cory comes up with. However, I
think it is possible to draft the demo requirements for the above
without going into too much technical detail--other than saying how the
services are invoked and what the messages are. The rest of the service
contract can be specified in text, or at most, simple UML diagrams.
However, I think it's possible to do it without those too (even though I
regularly use UML in the course of my work, and choose it for
specifications and architecture descriptions reviewed by business
stakeholders where it is appropriate). (030)
What do you think? (031)
ast (032)
On Fri, 2006-04-07 at 19:00, Cory Casanave wrote:
> Andrew,
> Andrew,
> I do understand that the spec is "MDA Centered" and don't take any offense
> at all - what I presented is certainly vested in the way we do things. Of
> course, any way to present things has to be vested in something. I did
> consider this issue and suggest that paradigms for defining & presenting
> architectures are just as much a part of the demo as anything else, and as
> such alternative ways of presenting it should be encouraged.
>
> But, we have to start someplace - so my hope is that we can use this to nail
> down the demo and then if anyone wants to present it differently they are
> open to demo it. At that point (after we know what the demo is), then this
> essentially becomes part of our companies demo (and that of the OsEra
> Project) - of how to link business architecture to systems architecture
> using MDA.
>
> As for "ownership" in the "MDA meta model". First, there is no "MDA meta
> model", there are several of them - the one we use is the EDOC Enterprise
> Collaboration Architecture. There is no assumption of "ownership". There
> is an assumption that there is some form of community contract, the
> authority and evolution of that contract may come from any source. There is
> also no assumption (in this model) of an authority over the process - it is
> defined as a collaboration of the participants.
>
> In my opinion, it is important for the demo to express the expectations of
> the community in a way that is business centric and technology independent
> and to also specify specific technologies (such as web services) that
> realize that community. So if this direction is reasonable - I will make it
> more clear in the model & document that this is one way to express an SOA -
> others are welcome.
>
> -Cory
>
> > -----Original Message-----
> > From: soa-forum-bounces@xxxxxxxxxxxxxx [mailto:soa-forum-
> > bounces@xxxxxxxxxxxxxx] On Behalf Of Andrew S. Townley
> > Sent: Friday, April 07, 2006 12:24 PM
> > To: Service-Oriented Architecture CoP
> > Subject: Re: [soa-forum] Next level of detail for SOA demo
> >
> >
> > *steps to the center of the room*
> >
> > *looks at all of the OMG members/participants in the audience*
> >
> > *blinks*
> >
> > Um, Cory... Please don't hit me, but if I didn't know anything about
> > SOA, and I read the demo specification, I would come to the conclusion
> > that MDA and OMG-EDOC/ECA were a requirement for SOA.
> >
> > I can understand that you, as a contributor to OMG specifications, a
> > developer of tools that implement these specifications and as an OMG
> > participant, naturally are using the technologies you believe in to both
> > express and meet the requirements of the SOA demo, but would it be
> > possible to not have the implication that SOA requires OMG standards as
> > part of the document?
> >
> > Please don't get me wrong (really!). This isn't a personal thing at
> > all--I'm trying to be objective here. I would say the same thing if
> > someone from Microsoft had expressed the demo requirements in terms of
> > WCF/Indigo/BizTalk, or someone from IBM/Iona/etc. doing the same with
> > SCA, or any one of the JBI vendors.
> >
> > However, the OMG approach and specifications only represent one way to
> > implement SOA. Further, I almost see the MDA meta-model as being not in
> > the overall interests of the community (as peers), because it implies
> > that there is an "owner" of the community that dictates the business
> > processes. Also, if I understand it correctly, changes to the
> > meta-model require changes to the community's agent implementations. If
> > these agents are truly independent, they cannot be under the control of
> > or tied to the MDA meta-model, and, as such, the positioning of the MDA
> > meta-model as the foundation of the community implies a sense of control
> > that cannot exist in a community of peers. In this respect, MDA is no
> > better or worse than any other way of specifying the "rules" of the
> > community.
> >
> > It is for these reasons that I ask the question--not because you are
> > part of the OMG. If the intent of the demo is to prove MDA as a means
> > of implementing SOA, then fine, it's an MDA-does-SOA demo, and that's
> > cool. It also implies that the fate of the SOA rests solely on the
> > capabilities of MDA, so I would question if that's a position the OMG
> > really wants to be in. Maybe it does. Sometimes sink or swim is a good
> > strategy rather than just floating along with all the others.
> >
> > We have spent a lot of time talking about the community aspects, and all
> > of the people who have participated in these discussions have completely
> > latched on to the implicit association between a social community of
> > real people and the way community has been used to describe the
> > architecture of the SOA. I think this is a powerful metaphor and one
> > that is good for both this CoI and SOA in general, but it's a bit like
> > being pregnant--you can't stop half way. All successful communities
> > have leaders, but few have dictators.
> >
> > If we have to create a demo by the conference, then I have no big issue
> > with the technical details of the scenario other than those I have
> > already expressed. I have a couple of detail questions around the
> > necessity of the buyer to register with the registry and the implication
> > that SOA requires traditional middleware, but these are minor things.
> >
> > Once again, I am not attacking you, the OMG or anyone else. I just
> > think that if you believe in SOA and the things we've been discussing,
> > that this is more important than trying to claim that there is an
> > optimal set of tools and technologies to implement it. If vendors wish
> > to showcase products and specifications in the implementation of agents
> > within the community, I think that's fine, and a good opportunity for
> > them. However, I firmly believe that implicitly tying the success or
> > failure of SOA the concept to any particular vendor or technology is a
> > mistake and is not representative of the capabilities or the potential
> > of SOA the architectural style. In this respect, I think the demo
> > should be as objective as possible.
> >
> > I guess these comments will pretty-much rule me out of much else to do
> > with this effort, but I think they are necessary. SOA should stand or
> > not on its own merits, and I think it can do so. Yes, you have to
> > implement it and there are easier ways than others, but these are
> > implementation details, and I think they should not be represented as
> > anything otherwise.
> >
> > Thanks for listening and good luck,
> >
> > ast
> >
> > On Thu, 2006-04-06 at 21:18, Cory Casanave wrote:
> > > Enclosed expands on the "buyer/broker/manufacturer" example as a
> > > proposed SOA demo. This is not yet done but I want to get it out for
> > > feedback prior to doing much more work.
> > >
> >
> > >
> >
> > >
> >
> > > I actually started to do an SOA based on the records management spec
> > > and pulled-back because it seemed that much of it would just take to
> > > much interpretation and explanation - something we don't want for our
> > > first demo. There are a lot of special "archivist" words that are
> > > outside of most of our comfort zones. So I returned to the more
> > > simple (I know, some people think brain-dead) broker scenario. At
> > > least it is easy to understand and implement (a lot of systems should
> > > be able to be wrapped to implement & use the services it in minutes to
> > > hours, which makes good demo).
> > >
> >
> > >
> >
> > >
> >
> > > I had hoped to get this to the next level, complete with WSDL
> > > interfaces produced from the model, tested with simulation - but
> > > that's not done yet but will be in a couple of days (work keeps
> > > getting in the way). So the idea is to get at least 3 independent
> > > participants to implement components behind the services to bootstrap
> > > the community.
> > >
> >
> > >
> >
> > >
> >
> > > One area I really stripped-down is the SOA messages, they are very
> > > small, really just a demo. I looked at some of the industry schema,
> > > but they are so big it would be a bit hard to follow. So consider the
> > > tradeoff and provide feedback.
> > >
> >
> > >
> >
> > >
> >
> > > There was also a question as to what EDOC was and what a SOA community
> > > is - well, here it is. At least from one view.
> > >
> >
> > >
> >
> > >
> >
> > > The question was asked, again, as to the purpose of the demo - this is
> > > what I have:
> > >
> >
> > > The goals of this demonstration are;
> > >
> >
> > > * To provide a concrete example of how the SOA approach provides
> > > business value to a community
> > > * To provide confidence that the approach and technologies are
> > > real - secure, reliable, performing and practical.
> > > * To validate that independently developed applications can
> > > interoperate using SOA standards
> > >
> >
> > >
> >
> > >
> >
> > >
> >
> > > What I may want to add a non-goal; This is not a demo of what SOA may
> > > become or possible future approaches, this is to show how the best
> > > practice of SOA and supporting real technologies can provide business
> > > value RSN (Real Soon Now). Just the idea that independently developed
> > > systems can interoperate within an open community is a big deal to
> > > much of the business community, old hat to many of us, but still of
> > > great business value. So that is the essence of the business value.
> >
> > > We can then add to that all the great stuff we can do with our cool
> > > tools, infrastructures, ontology-stuff and approaches. If we can't,
> > > at lest, do this simple demo we should just go home.
> > >
> >
> > > Of course, the goals are also a part of the consensus process, your
> > > mileage may vary.
> > >
> >
> > >
> >
> > >
> >
> > > We need to get consensus on the scenario real soon, not the technology
> > > or SOA theory - but what the business intent of the SOA is. If not
> > > "broker" we need well developed alternatives ASAP.
> > >
> >
> > >
> >
> > >
> >
> > > Regards,
> > >
> >
> > > Cory Casanave
> > >
> >
> > >
> >
> > >
> >
> > > ______________________________________________________________________
> > > _________________________________________________________________
> > > 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
> > --
> >
> > Join me in Dubrovnik, Croatia on May 8-10th when I will be speaking at
> > InfoSeCon 2006. For more information, see www.infosecon.org.
> >
> > **************************************************************************
> > *************************
> > The information in this email is confidential and may be legally
> > privileged. Access to this email by anyone other than the intended
> > addressee is unauthorized. If you are not the intended recipient of this
> > message, any review, disclosure, copying, distribution, retention, or any
> > action taken or omitted to be taken in reliance on it is prohibited and
> > may be unlawful. If you are not the intended recipient, please reply to
> > or forward a copy of this message to the sender and delete the message,
> > any attachments, and any copies thereof from your system.
> > **************************************************************************
> > *************************
> > _________________________________________________________________
> > 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
>
> _________________________________________________________________
> 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
--
Join me in Dubrovnik, Croatia on May 8-10th when I will be speaking at
InfoSeCon 2006. For more information, see www.infosecon.org. (033)
***************************************************************************************************
The information in this email is confidential and may be legally privileged.
Access to this email by anyone other than the intended addressee is
unauthorized. If you are not the intended recipient of this message, any
review, disclosure, copying, distribution, retention, or any action taken or
omitted to be taken in reliance on it is prohibited and may be unlawful. If
you are not the intended recipient, please reply to or forward a copy of this
message to the sender and delete the message, any attachments, and any copies
thereof from your system.
***************************************************************************************************
_________________________________________________________________
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 (034)
|