Howard, I know we've had this discussion before here on groupstudy. I also
accept that there remains a lot of confusion of OSI terminology with PC
terminology. I.e. application on a PC versus OSI application.
That said, I am looking at the Certification Zone OSI paper. In it is the
statement that the application layer provides common services to specific
application programs, including file transfer, messaging ( not in the sense
of e-mail messages ), directory lookups, etc. ( I have paraphrased this a
bit, but I believe this is true to the original sentences )
If I have a Linux box, and on that Linux box I am running RIP and OSPF, am I
not running user applications? Is the operation of those applications using
services at the OSI application layer? I.e. receiving messages, sending
messages, doing what might be arguably called "file transfers" etc. ? The
purpose of the user applications ( as opposed to the OSI application layer )
is to provide particular functions and distribute particular information,
is it not?
A router is a computer, after all. It's architecture is based on the Von
Neuman model. It has a CPU and memory and software that performs various
operations and functions. The is an operating system that handles the i/o
and integrated into that operating system are the various - dare I say it? -
functions or "applications"? ( as an aside, isn't this what Microsoft is
getting bonged for doing? :-> )
I agree that it is not the best use of time to spend countless hours
messages arguing how many angels can dance on the head of a pin.
But I do wonder if the code that results in a BGP table, or an OSPF database
might by some standards be considered "user applications" which in turn use
functions that are defined within the OSI application layer ( e.g. file
transfer - routing updates? - notification messages - hello packets?
LSA's? )
Yes, routing is a layer three function, but is it incorrect to say that
there are things going on at other layers that make the layer 3 functions
possible?
Chuck
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of
Howard C. Berkowitz
Sent: Saturday, November 11, 2000 6:53 AM
To: [EMAIL PROTECTED]
Subject: RE: routing protocols vs Tcp/ip model
>Stephen Skinner wrote, let me see if i have this correct...
>
>routing protocols do the following
>
>manage links
>exchange data
>route packet
>update each other
I'd agree with exchange data and update each other. Try approaching
this from a different direction. Can we agree that the endpoints of
an application layer communication are identified with application
layer identifiers (e.g., TCP port plus application-specific
information such as email user names), transport layer communications
have transport layer endpoint identifiers (address plus port), etc.?
The routers don't have application or transport identifiers (yes, I
know about BGP). Routing protocols control devices at the network
layer. The endpoints of communication between routers are IP
addresses, even if higher-layer mechanisms such as TCP are used for
actual information transfer.
One of the problems in understanding real-world protocols is that
they may do things in an indirect manner. Think of tunneling, where
you have to do a recursive lookup of two headers to determine the
real next hop. BGP is a network layer control/management protocol
that uses TCP for its transfer.
In contrast, SNMP does work at the application layer. Its endpoints
are identified by the combination of an IP address, a UDP port, and a
MIB identifier.
>
>what his instructor is saying is that some of these functions work
>at different "layers" of the OSI model...YES i know i shouldn`t be
>using that as a reference but for this argument i will...sorry...
>so tell me OSPF....what "layers" would the above function be compared to
>
>please .....i have a major headache now as i thought i knew it
>
>>From: "Howard C. Berkowitz" <[EMAIL PROTECTED]>
>> vs Tcp/ip model
>>Date: Sat, 11 Nov 2000 08:11:34 -0500
>>
>>Your instructor is one of the all-too-large group of people who try
>>to coerce things into a simplistic OSI model. Priscilla calls this
>>coercing protocols into OSI layers. It's really not the fault of OSI,
>>because there are documents that supplement the original model, such
>>as the Rout(e)ing Framework, Internal Organization of the Network
>>Layer, Management Annex, etc.
>>
>>The OSI stack principally was drawn to show how standard
>>communications service user applications, which run on top of the
>>service interface to the application layer. Management was something
>>of an afterthought, and what is called system management -- think
>>SNMP, or the OSI rough equivalent, CMIP -- does indeed involve an
>>application layer protocol and a management application above it.
>>
>>Routing, error notification, etc., are considered layer management.
>>There is nothing "above" them; they are part of the infrastructure
>>for a given layer. So,all of them are logically layer 3.
>>
>>The issue of the mechanism they use to transfer information between
>>them is independent of the layer they manage. In Chuck's table
>>below, EIGRP and OSPF do have transport functions that are part of
>>their own design--which have a TCP-like flavor. For that matter, ISIS
>>runs directly over data link.
>>
>>>Recently an instructor in a class I was taking said something I found
>>>interesting. I hope I can do justice to his words.
>>>
>>>Network layer: IP IP IP
>>>Transport layer: TCP UDP
>>>Application layer: BGP RIP EIGRP, OSPF, IGRP
>>>
>>>In other words, he suggested that routing protocols are application
layer,
>>>and use the chosen transport or network layer protocols to communicate.
>>>Other reading I have done kinda says this in other ways. RIP uses UDP
port
>>>500. BGP, as we all know, uses TCP.
>>>
>>>Does this make sense?
>>>
>>>Chuck
>>
>>A post I made yesterday might help:
>>At 4:16 PM -0500 11/10/2000, Howard C. Berkowitz wrote:
>>>At 12:22 PM -0800 11/10/2000, Julian Eccli wrote:
>>>>Does anyone know the definition of Control Plane from a generic
>>>>routing protocol
>>>>standpoint? Is it the same definition as in ATM? I have heard
>>>>references to
>>>>control planes in various talks but they were not specific to ATM.
>>>>
>>>
>>>Unfortunately, it isn't as well-specified in IP routing as in the
>>>B-ISDN/ATM architecture. Many IP discussions merge what that
>>>architecture calls the control and management plane.
>>>
>>>Personally, I think merging the two is rather unfortunate. In IP
>>>networks, I consider control plane protocols those that are used for
>>>signaling between hosts and ingress/egress routers. Examples: ARP,
>>>IGMP. Another way to think about them is that they serve a
>>>user-to-network role.
>>>
>>>I consider pure management plane protocols to those used between
>>>routers: BGP, OSPF, EIGRP, RIP, etc. Arguably, these have a
>>>network-to-network role.
>>>
>>>There are protocols that don't neatly fit, such as RSVP and ICMP. I
>>>suppose they are control plane when host initiated and management
>>>plane when router initiated, but that doesn't always work and is
>>>ugly anyway.
--
"What Problem are you trying to solve?"
***send Cisco questions to the list, so all can benefit -- not
directly to me***
Howard C. Berkowitz [EMAIL PROTECTED]
Technical Director, CertificationZone.com
Senior Product Manager, Carrier Packet Solutions, NortelNetworks (for ID
only)
but Cisco stockholder!
"retired" Certified Cisco Systems Instructor (CID) #93005
_________________________________
FAQ, list archives, and subscription info:
http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
