So, an SOA service can work on its own w/o an invocation from a
consumer. Here is why I think it is possible:
 
 - due to the business orientation of SOA, we have to distinguish between
the action intent, trigger and request. Intension does not necessary lead to
the action but forms the “shape” of potential action. Trigger may lead to
action and request leads to action. 
 
Thus, we can easily imagine a News Agency that generates some news
info and broadcasts it in the absence of consumers. This business model is
based on soliciting demand and real consumers. While there may be no requests
at the beginning, there is a trigger – the business idea and investments into
such News Agency.
 
Technically, this may be implemented, for instance, via a SOA
service, which sends ‘news’ messages to the MOM Destination for broadcasting
via Topic model. Consumers can subscribe and unsubscribe from the Topic and, in
some moments, there may be no subscribers at all.  
 
Why I think the Registry is inappropriate place for the service
initialization is this:
during the registry
process, the service is not initialized, the service provider just
announces service availability and related constraints (via Service
Description and interaction policies). Depending on the nature of the service
(business or infrastructural), potential consumer has to go through the
negotiation of the Service Contract or take the service as mandatory (Service
Description plays the role of Service Contract in this case). Only then
consumer can invoke the service, if needed.
 
Executing service w/o
requests (an event still leads to the request) is possible but, IMO, business
inefficient action in the majority of cases and it happens seldom.

- Michael 



________________________________
From: Udi Dahan <[email protected]>
To: [email protected]
Sent: Thursday, December 25, 2008 12:54:59 PM
Subject: RE: [service-orientated-architecture] Re: Yefim Natis is sure that 
"SOA is integration"


OK – could not a service at initialization register for these
callbacks from an infrastructure timer/scheduler service-y thing, thus causing
it to run periodically?
 
-- 
Udi Dahan - The Software Simplist
 
From:service-orientated- architecture@ yahoogroups. com
[mailto:service- orientated- architecture@ yahoogroups. com] On Behalf Of 
Michael
Poulin
Sent: 25 December 2008 14:34
To: service-orientated- architecture@ yahoogroups. com
Subject: Re: [service-orientated -architecture] Re: Yefim Natis is sure
that "SOA is integration"
 
Actually,
it cannot. If it is not an external consumer, it is an external process calls
for the service (Scheduler, autosys, Calendar, whatever). Service serves upon
a request. It is only waiters in the low- quality restaurants ask you 'is
everything OK?' every 10 minutes...
 
However,
request:response != 1:1, it may be 1:M (call-back, subscription, etc.)
 
- Michael
 

________________________________
 
From:Udi Dahan <thesoftwaresimplist @gmail.com>
To: service-orientated- architecture@ yahoogroups. com
Sent: Wednesday, December 24, 2008 9:42:37 PM
Subject: RE: [service-orientated -architecture] Re: Yefim Natis is sure
that "SOA is integration"
Could not a service run periodically,
without any external calls?
 
-- 
Udi Dahan - The Software
Simplist
 
From:service-orientated- architecture@ yahoogroups. com
[mailto:service- orientated- architecture@ yahoogroups. com] On Behalf Of 
Michael
Poulin
Sent: 23 December 2008 11:50
To: service-orientated- architecture@ yahoogroups. com
Subject: Re: [service-orientated -architecture] Re: Yefim Natis is sure that
"SOA is integration"
 
Here is a misunderstanding, Rob. Certainly, to run, a
service needs an external call. However,  self-contained
atomic SOA business service does not need any other business
services to provide its RWE; it does not know about outside world of services.
In contrast, an aggregate service - does.
- Michael
 

________________________________
 
From:Rob
Eamon <rea...@cableone. net>
To: service-orientated- architecture@ yahoogroups. com
Sent: Tuesday, December 23, 2008 4:01:52 AM
Subject: [service-orientated -architecture] Re: Yefim Natis is sure that
"SOA is integration"
--- In service-orientated- architecture@ yahoogroups. com, Michael 
Poulin <m3pou...@.. .> wrote:
>
> Integration with what?
> 
> Assume, we have a self-contained atomic SOA business service. It 
> does not need anything outside its boundaries to perform announced 
> business functionality and provide for the RWE. 

Oh yes it does. A service all by itself does nothing. Without some 
external stimulus, it does nothing at all.

Something outside of the service must call it via one of the exposed 
interfaces or the service will do nothing whatsoever.

> Certainly, it has to run in an execution environment and it 
> integrates with it. However, it does not integrate with an 
> orchestration or a process that uses it because it perfectly 
> functions alone. 

The orchestration or process integrates with the service by 
connecting to and invoking one of the service's exposed interfaces.

> Invocation of such service does not generate any new business value.

Invocation of a service is required to generate any busines value. A 
service uninvoked is a useless pile of bits.

> As I responded to Rob, if a SOA service does not have an interface 
> for particular type of communication channel, does it " have 
> intrinsic seamless integration capabilities" ?

No. But then how did the architect miss that channel? But for the 
channels/interfaces that are not missing, the system has intrinsic 
integration capabilities. 

> So, integration is just a link system between participants (may be 
> one link for 2 participants) . Connecting those participants we do 
> not create a SOA system, or we do?

Connecting participants creates an integration. Creating an "SOA 
system" (a term I loathe) requires creating service providers with 
separately standing interfaces that are invoked within an execution 
context by service consumers. Consumers integrate with providers via 
the agreed upon interfaces.

-Rob
 
No virus found in this incoming message.
Checked by AVG - http://www.avg. com
Version: 8.0.176 / Virus Database: 270.10.0/1861 - Release Date: 22/12/2008
11:23
  
No virus
found in this incoming message.
Checked by AVG - http://www.avg. com
Version: 8.0.176 / Virus Database: 270.10.0/1861 - Release Date: 24/12/2008
11:49 


      

Reply via email to