Re: Resolution of Web Services Architecture Issues 27 and 28

> > Is it too late to add a normative input?
>
> Figure out how the OSI model normatively relates to the Internet / Web as
> actually deployed and we'll be happy to apply that to Web services :-)
>
> The efforts I've seen to do this involve a lot more hand waving than
formal
> mapping or analyses of how OSI concepts really and truly drove the
evolution
> of the 'Net.  That's not to say it's not a good "reference model", just
that
> it's not the architecture of what we actually use.
>

I think I understand.  You mean "normative" in the sense that the OSI would
be a prescription for how a networked product is built.

What I meant was more like that when networking stacks are referenced in
the WSA as "layer this" and "layer that", that those references would be,
by prescription*, a reference to some known model or another.  This would,
for example, prevent someone from saying "network layer" and meaning
the entire of what is encompassed in the OSI model.  And as you may know,
I'm not making this example up.

Another alternative is not to reference networking layers at all.  How
viable is that?

* Similar to how the WSA Requirements are requirements for an architecture,
not for a networked product itself, this kind of prescription would apply at
the
same level of abstraction.

I'll leave it there.  I don't think this group relishes another long
discussion
about this.

Walden

Received on Friday, 4 April 2003 09:26:08 UTC