[RINA] Question for all of you

John Day jeanjour at comcast.net
Sat Oct 2 21:47:51 EDT 2010


Thanks guys,

This has been a help.  I should have given you more to go on, but I 
didn't want to prejudice the reply any more than I had to.

 From your responses and others off list, I think I am getting a 
handle on where the discrepancy is in the Internet mindset. 
Primarily it probably starts with their focus on just the data 
transfer part of the layer and the emphasis on the non-issue of soft 
state.  What I had perceived was a view that can be characterized as: 
In the Internet, if an something requests service from a layer, it is 
the job of the layer in that system to figure out how to respond, 
rather than the function of the members of the layer.

That is part of it. The other part is that the Internet never had a 
concept of "layer management" as integral and necessary part of the 
layer.  This appears immediately in the earliest drafts of the OSI 
model, as does the idea of the different phases (for them) 
enrollment, establishment and data transfer). The first is indicated 
by the confusion of where routing resided.  In OSI it was always 
clear, it was layer management.  The last was indicated when Joe 
Touch told me enrollment was a RINA concept and I had to point out to 
him that it was 30+ years old in networking and pre-dated RINA by a 
long time.

They have also not understood the importance of scope.  Everyone of 
the so-called "cross-layer" papers fails because it does not 
understand scope.  The application being able to tell the physical 
layer precisely what it needs is mostly useless because that physical 
layer (or data link layer) is the only one between the source and 
destination.

So basically, the reason the Internet has always had such a hard 
problem with layers is that they never understood them.

A Layer is all the things you guys said it was but most importantly 
it is a cooperating collection of processes that provide IPC.

Thoughts?
John

At 13:05 -0400 2010/10/02, Yuefeng Wang wrote:
>Hi John,
>I think a layer is to provide unique services to the layer above it 
>by taking certain actions within this layer, also by using the 
>services from the layer directly below it.
>
>Regards,
>             Yuefeng
>
>Quoting John Day <jeanjour at comcast.net>:
>
>>Lately, I have had inklings from discussions on various IETF lists
>>that what I thought a layer was may be at odds with what much of the
>>Internet thinks a layer is.
>>
>>So I thought I would ask all of you, what it was you were taught a
>>layer was?  Could you define or otherwise characterize the concept
>>of"layer"?  Not layer in RINA, but the concept of layer as you
>>understood it prior to RINA (in case knowledge of RINA has changed
>>your concept of layer).  Of course it would be interesting to know
>>whether or not it has.
>>
>>This is not a trick question.  I suspect that there is a subtle
>>difference and this is the only way I can think to test the
>>hypothesis.
>>
>>Take care,
>>John
>>_______________________________________________
>>RINA mailing list
>>RINA at cs.bu.edu
>>http://cs-mailman.bu.edu/mailman/listinfo/rina



More information about the RINA mailing list