User talk:Mcnuttj

Routing management protocols

Please take this to the OSI model discussion page, where I just wrote a note, with citations, on why routing protocols are, indeed, network layer. I speak here from direct experience both as an OSI protocol architect and as a routing protocol developer. Perhaps you will want to hear what I am saying from other people.

Again, do you have authoritative sources that document your assumption about the placement of these protocols in the OSI model? I have quoted several.

When you say "Routing protocols, such as RIP, OSPF, and IS-IS are applications that allow routers (hosts) to trade information with one another. They do not create or modify existing addressing schemes" no. See multiprotocol BGP, nor can they be used independently of the network layers below them.", there is no network layer below them.

This is incorrect. Two routers using RIP or OSPF or IS-IS CANNOT COMMUNICATE WITH EACH OTHER until there is an active IP network in place.

"ARP, unfortunately, confuses matters because it is related to IP, but has its own Ethertype (which would seem to classify it as layer 3). ARP is *also* arguably a layer 7 protocol, but is usually classified as layer 3 because it is usually very difficult (though possible) to use IP without ARP." ISO 8648 defines this mapping function as part of the network layer, and with more protocols than IEEE MAC addresses Howard C. Berkowitz (talk) 18:16, 13 December 2007 (UTC)[reply]

This point is undisputed. ARP, though not strictly necessary to use IP, is only used when IP is in use and is part of the interface between IP and the layers below it.

Howard, you, as a researcher, have fallen into a trap that I find extremely prevalent among research and academic folks. You use your credentials as evidence for why you are correct. This is completely invalid. Your credentials lend weight to your opinion only in the sense that you probably have an opinion worth listening to. It makes your opinion no more or less correct. Someone with no training whatsoever can still prove that you are wrong.

Secondly, your citations are largely of ISO and OSI documents, which you freely admit do not strictly apply to the IP protocols, which are defined by RFC and then later redefined - somewhat on the fly - through real-world use.

There are two methods of classifying routing protocols: as applications or as network-layer protocols. Since, from an OSI standpoint, both classifications are arbitrary, one would think that the logical descriptions I have provided regarding how these protocols work and where their responsibilities lie (and do NOT lie) would be sufficient, yet you stubbornly cling to your equally arbitrary, misleading, counter-intuitive, and unhelpful model. This is why you wind up doing so much work undoing the edits of everyone who comes along to correct it. It doesn't make good sense.

I suggest you take a good look at your model and decide whether you defend it because you really believe that it is the most correct, or whether it is merely because that's what you decided a long time ago and you aren't going to change your mind. Justin McNutt