Services
Hello all,
In five days we will have a 90 minute discussion on "service". We should
first of all decide on a definition and then see how this concept fits
into the existing schema.
I will now propose three (rather crude) definitions, and we should
decide if we want to tag one of these as "service". Of course, this does
not exclude the possibility to still include one of the other
definitions in the schema, albeit under a different name.
1. A service is the act of providing something,
in particular of using the data plane
2. A service is the act of offering something,
in particular of offering the use of the data plane
3. A service is the act of providing information,
in particular providing control-plane services.
Examples of 1. are:
- a data connection
- an adaptation connection
- a data storage
Examples of 2. are:
- offering a data connection
- describing an adaptation capability
- offering a storage service
Examples of 3. are:
- providing a lookup service (yellow pages)
- providing a topology description service
- an authentication service
At this moment, the definition of the Service object in the schema draft
is as follows:
"A Service object describes the capabilities of node in terms of
switching and adaptation between different technology layers."
To me, this seems as a subset of definition 2.
I recall that Martin claimed that he like to remove the specific
references of "switching" and "adaptation" from the definition (those
were only examples of a service). My thinking is similar to this: I like
a "service" to be any offering by the data plane, but not to extend this
definition to describe a specific offering (for that, there is the Link
concept, Path concept and Relation concept), neither to extend it to the
control plane service (that is so generic that I think it is
out-of-scope for NML, and in that case, I tend to prefer either an
extension later, or a reference some other work.)
Regards,
Freek
