>At 09:10 AM 12/14/01 -0500, Howard C. Berkowitz wrote: >> >To Chuck, I do not agree that the OSI model is "crap". Sometimes it can >>>add confusion, but for the most part it is fairly well defined. Also, no >>>one ever said TCP/IP follows the OSI model 100%. The concept of layering >>>is just very easy to see with the OSI model. TCP/IP generally has only >>>layers such as the application, network, transport, and physical. You >>>could throw in datalink in there I suppose. It certainly helps people >>>understand networks. Without the OSI model, it seems like a lot of random >>>musings. TCP/IP has a very clear transport and network and application >>>layer. >> >>Then how is it that the TCP/IP suite was developed before the OSI >>reference model was finished, largely by people that, at the time, >>were very hostile to the OSI work and vice versa. I was there at the >>time, and remember European delegates to ISO making comments like "we >>will never use protocols developed by the bomb-crazed American >>military." > >OSPF and ISIS have some similarities, yet one came out earlier than >the other. The similarities were taken as they were developed. >Just because it was not "set it stone" yet, does not mean it did not >exist.
Huh? > The standards were not atomically created, they are developed as >time goes on. Yes, I know. I was there. Were you? >As some parts were done, they probably took it and ran with it. >Plus it only seemed like a logical separation. It was mainly for >insulation of different layers (as you mentioned, good programming >practices) which created these divides. So, I think it is >reasonable to still say as a reference model, tcp/ip matched some >parts of the osi model. Who stole who, does not matter, they still >follow a similar layering for transport and network. And the >network layer for the most part, could care less what it runs over >as long as that is insulated from them. Quoting Winston Churchill when challenged by an indignant dowager, "Prime minister, you are drunk, drunk, very drunk, and disgustingly drunk," WSC replied, "Madam, I am indeed drunk, drunk, very drunk, and disgustingly drunk. And, Madam, you are ugly, ugly, very ugly, and disgustingly ugly. And further, Madam, in the morning, I shall be sober." In this case, what is definitive are specific ISO or IETF architectural documents about the placement of routing protocols, not textbooks that paraphrase standards or personal opinions without direct experience with the protocol standards development process. This is not to say that some of these issues are quite subtle. Believe me, these are being very actively examined in the IETF (or more specifically the Internet Research Task Force) as we look at the routing architecture to follow BGP. I'm coauthor of one early document in this area, http://www.ietf.org/internet-drafts/draft-davies-fdr-reqs-01.txt This paper has lots of holes in it and is intended to stimulate debate, which mostly is going on in the IRTF-RR working group. (routing research) >That seems very real in both practical and theory cases. That is >what tcp/ip does, that is what the osi model references. "similar" doesn't cut it when you are dealing with abstract models (in the sense of abstract algebra, which DOES apply to protocol design). > >> >To Jose, I feel they do not work at the network layer, and work at the >>>application layer. If it uses protocols, (EIGRP and OSPF) it uses IP RAW >>>which means it skipped the transport component, ultimately I still feel it >>>is at the application layer. >> >>In my sophomore year of high school, I _felt_ that a girl named Gail >>_should_ have reciprocated my affections and lust. She didn't. Just >>because, Carroll, you feel something, doesn't make it right. Ignoring >>the TCP/IP work, ISO says you are wrong in its "OSI Routeing >>Framework" document, in which routing protocols for layer N are >>defined as layer management protocols for and of layer N. The >>transport they use is irrelevant, because their payloads affect layer >>N directly. > >I did not mean I was being definitive. That is why I said I felt. >I was not sure, and told him my perspective since he was asking for >one. All of us seem to agree that the result / payload affects the >network layer. As long as we understand that part, I think that is >pretty good. Semantics aside. May I politely suggest that feelings aren't necessarily going to suppport success either with Cisco exams or operational practice, if there are more definitive references that contradict it? > >(That seems to be what Chuck is getting at. Screw the OSI model and >semantics, as long as we know what it is doing. However, I think >some people are not at that level to even know since we have no >semantics to work at all. The important key here is that we >understand it resides perhaps at another layer, but affects layer 3. >As opposed to it itself being at layer 3.) > >Anyway, I guess I am totally wrong on this. Sorry for wasting >everyone's time I will try not to respond anymore. I just felt that >it seemed like a good way to learn, and as a baseline, the OSI model >seemed ok, and I thought TCP/IP matched some of it. I guess it does >not match it at all, so learn the layering of tcp/ip elsewhere. > >-Carroll Kong Then you have learned something important. Don't try to coerce TCP/IP into OSI, or for that matter into SNA. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=29209&t=29139 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
