Re: [soa-rm] Definition of Service:From concall

[Ken Laskey <klaskey@mitre.org>]

Been meaning to reply to this for a while.  See comments below.

Ken

At 04:51 PM 8/10/2005, Sally St. Amand wrote:
>hi all
>On todays call I was trying to present my definition of what a 
>service is. If I had a whiteboard
>
>Org has          Translates              Digitized
>
>Need              Task           +          wire enable    =  Service
>Capability
>
>
>Start from the perspective that an org has a need (or is offering a 
>capability). That need gets translated into a task, eg get an 
>airline seat [which is a finer granularity; or an activity, make a 
>trip reservation, which is more complex] The task has



>to have technical details added to allow it to be provided and/or consumed.

I would complete the sentence as:  The task has defining parameters 
that provide indications of the capabilities needed to accomplish the 
task.  The task is neither provided or consumed but represents a real 
world effect that is desired.

>The end result is a service



>in the context of our work for SOA.
>

I would complete the sentence as: The possible end result is a 
service found to invoke the capabilities needed to accomplish the 
task.  This may require additional parameters or other information 
required by the service and which go beyond the original task 
definition.  For example, the service requires my email address but 
there was no need for that to be part of the original task definition.

>A service is based on the functionality it is providing. That 
>functionality is not just the business (or mission) task/activity 
>but the digital requirements, eg encryption, endpoints. A service is 
>multi faceted.
>

Generally agree .

> From this perspective shoe shining and food delivery are not services.

I could say there are underlying capabilities (e.g. shoe shining 
equipment exists, someone knows how to use the equipment) and the 
service delivery mechanism is that someone manning a shoeshine 
stand.  This is not a service of interest in the SOA sense but if 
pressed, I could identify a service interface, a service invocation 
protocol, and the policies and constraints upon which the provider 
and consumer must agree.

>So a logical conclusion is that 'services' are not absolute.
>

Ignoring for the moment dynamic composition, what defines a service 
is fairly stable -- a fast time constant for change could be on the 
order of days during early deployment but not the order of 
minutes.  The tasking and use of services is not absolute.

>Sally
>

--
      ---------------------------------------------------------------------------------
   /   Ken 
Laskey                                                                \
  |    MITRE Corporation, M/S H305    phone:  703-983-7934   |
  |    7515 Colshire Drive                    fax:      703-983-1379   |
   \   McLean VA 22102-7508                                              /
     ----------------------------------------------------------------------------------