Re: [soa-rm-ra] interaction question [was: DM2 + SoaML

[mpoulin@usa.com]

Unfortunately, I did not participate in the discussion mentioned by Boris but this topic is one of my subjects of research.

In particular, orchestrated or simple service - this does not matter for the consumers, it is the service. For development (including management governance and even testing) they are different but both adhere to the Principles of Service Orientation. Plus, as I noticed in my BLOG, even Thomas Erl in his comments to the @"atomic" services, which are stateless and complete fairly quickly@ said that statelessness was never meant as a condition, it came out just as an implementation reality; the service must manage its state based on the business task/condition (Thomas said); everything else is the technology/implementation tricks. This is why I had to review the Principles of Service Orientation () making them more technology independent. Also, the term "atomic" is says nothing that the service does not use other services, i.e. it is not divided further in the service decomposition.  

Some aggregate/orchestrated services can run faster than the "atomic" ones.Based on what I said before, a SOA service can engage a Workflow as its resources (which is not a service) and sit and wait when the humans respond to the Workflow actions. This is the implementation detail that does not change the nature of the atomic service. A little bit different story when the orchestration uses a long-running transaction between services. Even in this case, we cannot guarantee that all orchestrated services are "atomic", some of them may be orchestrations of their own. Composability it the major SO principle and the power of SOA.

- Michael

-----Original Message-----
From: Lublinsky, Boris <boris.lublinsky@navteq.com>
To: Ken Laskey <klaskey@mitre.org>; soa-rm-ra@lists.oasis-open.org RA <soa-rm-ra@lists.oasis-open.org>
Sent: Sun, Nov 22, 2009 3:52 pm
Subject: RE: [soa-rm-ra] interaction question [was: DM2 + SoaML]

Ken,

I am inclined to think that this exchange is highly related to the conversation that we were having last wensday about services interactions and orchestrations.

This is a question of whether we differentiate between "atomic" services, which are stateless and complete fairly quickly and "orchestrated" services (aka business processes), which support conversational semantics and can last for a very long period of time.

In my mind, those are both types of services, but are sufficiently different, not only technologically, but design wise, that they can and should be discussed and defined differently.

I am for separating of real world effects - results - from  value exchanged between participants when a service is enacted - means.

 

---

From: Ken Laskey [klaskey@mitre.org]
Sent: Sunday, November 22, 2009 9:18 AM
To: soa-rm-ra@lists.oasis-open.org RA
Subject: [soa-rm-ra] interaction question [was: DM2 + SoaML]

This is an exchange with Cory Casanave that is part of the DM2 and SoaML work.  Start from the bottom, as usual, and then consider the following.

Do we need to say more about service interactions where the real world effects can occur long after the passing of messages has completed?  I'm not an expert in the definition of transactions but I don't think of the example I give below (requesting a copy of an explanation of benefits and later paying a fee) as just one "classic" transaction.  If I don't pay the bill and it goes to collection and I dispute it on constitutional terms that get appealed to the Supreme Court, is that one transaction?  Assuming SOA services, when do interactions begin and end?

Ken

Begin forwarded message:

From: Cory Casanave <cory-c@modeldriven.com>

Date: November 21, 2009 11:43:57 PM EST

To: Cory Casanave <cory-c@modeldriven.com>, "Laskey, Ken" <klaskey@mitre.org>

Cc: "Dandashi, Fatma" <dandashi@mitre.org>, "Bock, Conrad" <conrad.bock@nist.gov>, Dave McDaniel <davem@silverbulletinc.com>

Subject: RE: DM2 + SoaML

Ken,
Also, I think it should also be said that SoaML services may be long
lived (months or years), complex and multi directional.  I would expect
the final "months later" action or payment that was the satisfaction of
the service obligation to also be part of the service contract and real
world effects.
I think this would be the same in the Oasis RM.  True?
-Cory

-----Original Message-----
From: Cory Casanave
Sent: Saturday, November 21, 2009 11:35 PM
To: 'Ken Laskey'
Cc: Dandashi, Fatma; Bock, Conrad; Dave McDaniel
Subject: RE: DM2 + SoaML

Ken,
Yes it does, or at least could - if modeled that way.  The value
exchanged is then an obligation for the future action.
-Cory

-----Original Message-----
From: Ken Laskey [mailto:klaskey@mitre.org]
Sent: Saturday, November 21, 2009 5:42 PM
To: Cory Casanave
Cc: Dandashi, Fatma; Bock, Conrad; Dave McDaniel
Subject: Re: DM2 + SoaML

Note, with respect to real world effect, these are (OASIS definition)  
the observable changes in the world that result from the interaction.   
The participants may find value in the real world effects -- it is  
likely if they were willing to engage in the interaction and the real  
world effects were sufficiently described so there weren't surprises.   
However, this is only "value exchanged between participants when a  
service is enacted" if you consider this value exchange to include all  
real world effects that may occur out of band later.  For example, I  
may request an explanation of benefits that will not be processed for  
several days.  After the request is processed, I will be billed a  
service charge and I will be expected to pay that service charge  
within some elapsed time.  The interaction set a process in motion   
that provides the explanation of benefits and a service payment but  
there are things that will occur after the interaction itself is long  
complete.

So, does the idea of value exchange cover from when the interaction is  
initiated to when all actions deriving from that interaction are  
complete, even if this is months later?

Ken

On Nov 21, 2009, at 3:12 PM, Cory Casanave wrote:

Also, on the definitional fragments in your document:

Collaboration:

- This word sense would be better: 1 : to work jointly with others or

together especially in an intellectual endeavor

- A collaboration in UML and SoaML specifies how participants will  

work

together for a purpose.

Real world effect - the real world effect in SoaML is the value

exchanged between participants when a service is enacted.

Port - the word sense is clearly not the one intended.  A port is an

interaction point.

---

However, I don't think the definition fragments or Webster's will  

get us

very far, we should be using the full definitions of the model  

concepts

from the specification and the examples which illustrate these.

-Cory

-----------------------------------------------------------------------------

Ken Laskey

MITRE Corporation, M/S H305      phone: 703-983-7934

7515 Colshire Drive                         fax:       703-983-1379

McLean VA 22102-7508

---

The information contained in this communication may be CONFIDENTIAL and is intended only for the use of the recipient(s) named above. If you are not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication, or any of its contents, is strictly prohibited. If you have received this communication in error, please notify the sender and delete/destroy the original message and any copy of it from your computer or paper files.