RIP and OSPF, which layer?

networker050184

SdotLow wrote: »

So BGP and RIP are themselves applications???

Depends who you are asking, me or Foresaken.

SdotLow

creamy_stew wrote: »

Let's say I have "hub" router with static routes pointing the networks in my organization to the proper next hop.

Every 10 minutes or so, I have a script that pushes route changes to this router via tftp. This is certainly completely transparent to the end user.

Is that,then, a L3 function?

It would be as much of a L3 function as manually typing it in through Telnet would be. The user is still setting up the application to alter the router through application layer protocols.

A static route changing on a router isn't a routing protocol, I wouldn't think.

MrRyte

Maybe my post was a little blunt; but I still can't see how it can be considered an application......

The only time that a person would interact (either by console or telnet)with the routing protocols is when they are configuring them or troubleshooting them. Otherwise; they are supposed to be working behind the scenes ensuring the correct routing of packets.

Turgon

networker050184 wrote: »

The end user doesn't, but someone configures them, apply policy etc. The same way you interact with any other application.

Its really a debatable question. Some people say application, some say network. RIP for example uses UDP for communication. Does that make it an application? Guess its up to interpretation.

But UDP is connectionless transport layer 4, so is a routing protocol a transport layer mechanism? jk

OSI layers can become blurred and a lot of fun to be had with the circular debates!

SdotLow

MrRyte wrote: »

The only time that a person would interact (either by console or telnet)with the routing protocols is when they are configuring them or troubleshooting them. Otherwise; they are supposed to be working behind the scenes ensuring the correct routing of packets.

I would say that a even if you're using telnet to configure a router, you're not interacting with a layer 3 protocol so to speak. The protocol is what it (the protocol) does, propagating routing tables and assigning packets to x/y/z route.

Network: Can you elaborate on why BGP or RIP would be considered an application?

Turgon

SdotLow wrote: »

It would be as much of a L3 function as manually typing it in through Telnet would be. The user is still setting up the application to alter the router through application layer protocols.

A static route changing on a router isn't a routing protocol, I wouldn't think.

A static route is not a routing protocol.

SdotLow

Turgon wrote: »

But UDP is connectionless transport layer 4, so is a routing protocol a transport layer mechanism? jk

OSI layers can become blurred and a lot of fun to be had with the circular debates!

I've read that they get blurred from interaction, from say layer 2 to 3 in a switch or what have you. But how does it get blurred from layer 3 to 7?

I'm honestly trying to wrap my head around what I'm missing here. Forsaken had me thinking I wasn't crazy but then Turgon jumps in and mixes it up again.

Turgon

networker050184 wrote: »

Where would you put LDP on the OSI model? Not trying to be difficult, curious what your opinion is on that?

Good one. A lot of literature passes it off as layer 2.5

Turgon

SdotLow wrote: »

I've read that they get blurred from interaction, from say layer 2 to 3 in a switch or what have you. But how does it get blurred from layer 3 to 7?

I'm honestly trying to wrap my head around what I'm missing here. Forsaken had me thinking I wasn't crazy but then Turgon jumps in and mixes it up again.

Lots of opportunities to blur higher up too. Coders

The TCP/IP Guide - OSI Reference Model Networking Layers, Sublayers and Layer Groupings


'In some areas, the layers are so closely related that the lines between them become blurry. This is particularly the case when looking at the higher layers; many technologies implement two or even all three of these layers, which is another reason why I feel they best belong in a group together. One important reason why the distinctions between layers five through seven are blurry is that the TCP/IP protocols are based on the TCP/IP model, which combines the functions of layers five through seven in a single, thick layer.'

Sorry to mess with your head!

networker050184

SdotLow wrote: »

I would say that a even if you're using telnet to configure a router, you're not interacting with a layer 3 protocol so to speak. The protocol is what it (the protocol) does, propagating routing tables and assigning packets to x/y/z route.

The routing protocol does not assign packets to routes. All the routing protocol does is exchange the information and run it against a best path selection algorithm. The forwarding process then uses that information to actually route the traffic.

I'd say BGP is an application because it take NLRI (network layer reach ability information), runs a best path selection algorithm and then offers that information to the router to form the routing table. The NLRI can be straight up IPv4, VPNv4 etc.

Same with RIP. Its just an application used to exchange information and run an algorithm against it. The router uses that output to build its routing table. The routing protocol itself isn't doing any routing function on the router.

My example with LDP is the same way. Its an application designed to transfer MPLS label information between routers. Just because it happens to carry labels doesn't mean it operates on the same layer as the labels used for forwarding.

networker050184

Turgon wrote: »

Lots of opportunities to blur higher up too. Coders

The TCP/IP Guide - OSI Reference Model Networking Layers, Sublayers and Layer Groupings


'In some areas, the layers are so closely related that the lines between them become blurry. This is particularly the case when looking at the higher layers; many technologies implement two or even all three of these layers, which is another reason why I feel they best belong in a group together. One important reason why the distinctions between layers five through seven are blurry is that the TCP/IP protocols are based on the TCP/IP model, which combines the functions of layers five through seven in a single, thick layer.'

Sorry to mess with your head!

Agreed. Its not as straight forward as it would seem if you look deeper into it and the layers blur together haha.

Turgon

networker050184 wrote: »

The routing protocol does not assign packets to routes. All the routing protocol does is exchange the information and run it against a best path selection algorithm. The forwarding process then uses that information to actually route the traffic.

Pretty much. Routing protocols offer routers tables to query by a process on the router. Essentially you are determining which exit interface on the device to use to get to an IP destination that may be close by or many hops away. Once that is determined the correct layer 2 encapsulation type has its way. Ethernet, serial, frame relay, whatever. Reading about RIB and FIB might help anyone struggling with some of this.

CaySpekko

To really understand the OSI stack you need to understand network programming.

One of the reasons for the OSI model to help programmers and manufactures have a model to follow when implementing network interactions, so that networks are interoperable without to much proprietary configuration.

OSPF and RIP advertisements and hellos are at most encasulated into a layer 3 packet, then the packet is pushed into the MAC layer, where it's encapsulated with a mac header, then pushed on the physical wire, then to the other router, when de-encasulates the mac, sees the ip header, strips that off. The data left over is sent straight away to the OSPF process without transending other network layers (the router knows it's ospf because it came from multicast 224.0.0.5 and 224.0.0.6). At most it uses layer 3. Sure the OSPF software running on the router is an application, but it's application is internal to the router, and not used in the network stack layers 4 5 6 or 7.

Telnet, is used to test all the way upto layer 7, and this is a good test question. If telnet is successful, then you know the entire 7 layer stack is working properly.

The flow is somewhat as follows:
Application: The application will request to the OS that it needs a socket for communication over the network.

Presentation: the telnet app determines that it's going to send data over the link using plain text, probably using an array as a data structure.

Session: Telnet argues with the other end and establish a method for communication, the actual datagram isn't modified in this layer, but the app will send communication requests out of the socket, and wait for a response back, before it tries to forward data.

Transport: telnet uses TCP port 23 the layer 4 datagram encapsulates the raw telnet data

Network: IP addresses encasulate the layer 4 datagram (it's all part of the ip header)

Datalink: mac address encapsulate the layer 3 packet

Physical: the mac frame is sent out the wire as bits

Even if you run OSPF on Windows Server, it'll listen on 224.0.0.5 and 224.0.0.6, and send the data from the IP packet straight to it's OSPF process, and doesn't have to do all the higher layer networking stuff.

Turgon

CaySpekko wrote: »

To really understand the OSI stack you need to understand network programming.

One of the reasons for the OSI model to help programmers and manufactures have a model to follow when implementing network interactions, so that networks are interoperable without to much proprietary configuration.

OSPF and RIP advertisements and hellos are at most encasulated into a layer 3 packet, then the packet is pushed into the MAC layer, where it's encapsulated with a mac header, then pushed on the physical wire, then to the other router, when de-encasulates the mac, sees the ip header, strips that off. The data left over is sent straight away to the OSPF process without transending other network layers (the router knows it's ospf because it came from multicast 224.0.0.5 and 224.0.0.6). At most it uses layer 3. Sure the OSPF software running on the router is an application, but it's application is internal to the router, and not used in the network stack layers 4 5 6 or 7.

Telnet, is used to test all the way upto layer 7, and this is a good test question. If telnet is successful, then you know the entire 7 layer stack is working properly.

The flow is somewhat as follows:
Application: The application will request to the OS that it needs a socket for communication over the network.

Presentation: the telnet app determines that it's going to send data over the link using plain text, probably using an array as a data structure.

Session: Telnet argues with the other end and establish a method for communication, the actual datagram isn't modified in this layer, but the app will send communication requests out of the socket, and wait for a response back, before it tries to forward data.

Transport: telnet uses TCP port 23 the layer 4 datagram encapsulates the raw telnet data

Network: IP addresses encasulate the layer 4 datagram (it's all part of the ip header)

Datalink: mac address encapsulate the layer 3 packet

Physical: the mac frame is sent out the wire as bits

Even if you run OSPF on Windows Server, it'll listen on 224.0.0.5 and 224.0.0.6, and send the data from the IP packet straight to it's OSPF process, and doesn't have to do all the higher layer networking stuff.

Very good point. I supported FX trading app developers and testors for a number of years. For networkers I recommend books on windows sockets programming and Stevens. Once you get into heavy weight real time trading environments it pays to know a good deal.

networker050184

Good explanation CaySpekko. What are your thoughts on something like BGP or LDP that does establish connections etc? Thanks for the input!

MrXpert

SdotLow wrote: »

You didn't look very hard. You could have gotten the answer on Wikipedia.

An application layer protocol is something like Telnet, HTTP, or FTP. Something communicates with an application, so to speak. OSPF and RIP are never seen by an application, or the application layer.

With that being said, you're posting in the CCENT/CCNA forums about a very easy question that can be found in your study book(s). I'd suggest looking there.

Put it this way "I looked hard enough to see it was worthy of debate"
i have seen at least two different sources stating different things including a discussion about it on the cisco forums website between CCNP and CCNAs who also appear to be not so sure about it

MrXpert

hiddenknight821 wrote: »

You got to be joking. It's a silly debate and you're asking us for help?! This is not some tag-team wrestling. Looks like both of you need go back to the books and review on OSI layer.

I'm surprised at your answer considering you're a ccent and i am not. is it such a silly debate? then why so many posts about it?
I personally tend to question things when different sources say different things.Looks like I am not the only one needing to rehit the books as you suggested

MrXpert

MrRyte wrote: »

And HOW EXACTLY does the end user interact or use RIP, OSPF or any other routing protocol?

I think that the answer is obvious....

crickey if you passed the ICND1 then i'll have no problem with it then

Forsaken_GA

networker050184 wrote: »

Where would you put LDP on the OSI model? Not trying to be difficult, curious what your opinion is on that?

Like you, I'd put it at layer 2.5 because that's where it would logically fit.

If I had to make a choice between layer 2 and layer 3, I'd put it at layer 3, as label lookup/forwarding is very similar to ip lookup/forwarding, given that labels are normally tied to a prefix, which means they're tied to the routing table.

As for the rest of it, CaySpekko made the arguments I would have (and better than I would have) so I'll let it alone

Forsaken_GA

CaySpekko wrote: »

Sure the OSPF software running on the router is an application, but it's application is internal to the router, and not used in the network stack layers 4 5 6 or 7.

This is where I think the key to the debate and confusion may lay. A protocol in and of itself is not necessarily an application, but the implementation of that protocol may be. Trying to apply operational details to a conceptual framework is a good way to generate a headache.

Chris_

I think forsaken has hit the nail on the head. The OSI model is a conceptual framework intended to assist in the demarcation of processes involved in communications. It is always going to be open to interpretation.

My own opinion is that routing protocols are applications that exist to facilitate the functions of layer 3. In my mind anything that processes and exchanges information that has payloads encapsulated within tcp/udp headers is leaning towards being an application (I know this doesn't apply to OSPF or eigrp but there us still a layer 4 exchange of sorts within the ip encapsulation)
Please don't attack me

hiddenknight821

MrXpert wrote: »

I'm surprised at your answer considering you're a ccent and i am not. is it such a silly debate? then why so many posts about it?
I personally tend to question things when different sources say different things.Looks like I am not the only one needing to rehit the books as you suggested

No need to reply here since I apparently just learned from this interesting debate as well as the others that you just refuted. Don't assume CCENT will teach you this. Not even the CCNA made it clear as well. If it wasn't for the experienced CCNP folks here, you would probably still be wrestling around with your friends. Anyway, no one wins in this debate since there is no definite answer yet. So far CaySpekko made so much sense out of this whole discussion

CaySpekko

networker050184 wrote: »

Good explanation CaySpekko. What are your thoughts on something like BGP or LDP that does establish connections etc? Thanks for the input!

I'm assuming you mean Border Gateway Protocol and Label Distribution Protocol, both of which according to RFC are application layer protocols on the TCP/IP model. I actually don't know a whole lot about those two protocols, but a great resource is: IP protocol suite

One thing I'd like to point out is that if you look at a BGP packet you'll run into this:

MAC header | IP header | TCP header | BGP header | Data ...

Notice that the BGP header comes after the TCP header, so you can assume that it's part of the layers above the transport layer. Same goes with LDP.

alan2308

Chris_ wrote: »

I think forsaken has hit the nail on the head. The OSI model is a conceptual framework intended to assist in the demarcation of processes involved in communications. It is always going to be open to interpretation.

My own opinion is that routing protocols are applications that exist to facilitate the functions of layer 3. In my mind anything that processes and exchanges information that has payloads encapsulated within tcp/udp headers is leaning towards being an application (I know this doesn't apply to OSPF or eigrp but there us still a layer 4 exchange of sorts within the ip encapsulation)
Please don't attack me

This is my thought on the matter. If it is encapsulated within TCP/UDP headers, then it has to be higher up the stack than TCP/UDP.

And I hate to start another debate, but OSI is more than a conceptual model. Cisco IOS still routes clns and family in 12.4 and perhaps later.

Forsaken_GA

alan2308 wrote: »

And I hate to start another debate, but OSI is more than a conceptual model. Cisco IOS still routes clns and family in 12.4 and perhaps later.

I'm not quite sure what point you're trying to make... clns is a layer 3 protocol, not the entire model itself.

The OSI model is a conceptual framework that illustrates how the information flows, not an implementation guideline.

pham0329

networker050184 wrote: »

The end user doesn't, but someone configures them, apply policy etc. The same way you interact with any other application.

Its really a debatable question. Some people say application, some say network. RIP for example uses UDP for communication. Does that make it an application? Guess its up to interpretation.

Going by that, wouldn't everything be at the application layer?

instant000

Key Concept: It is just as much a mistake to assign too much importance to the OSI Reference Model as too little. While the model defines a framework for understanding networks, not all networking components, protocols and technologies will necessarily fall into the model’s strict layering architecture. There are cases where trying to use the model to describe certain concepts can lead to less clarity rather than more. One should remember that the OSI model is a tool, and should be used accordingly.

Source: The TCP/IP Guide - How To Use The OSI Reference Model

alxx

networker050184 wrote: »

Damn Chrome spell checker! And no worry, not heated here. I think its a great discussion with valid points from both angles.



I'd say they are an application because they are a purpose built software used to fulfill a need for people. We could always static route everything right? Or do all of the SPF calculations manually and input the routes ourselves, but why? We can write an application that can do all of that for us. Hence routing protocols as applications that are ran by a router.

And whats the first application listed here?

TCP/IP model - Wikipedia, the free encyclopedia

But if a router is a layer 3 device doesn't that make the protocols and applications it uses also level/layer 3 ?

http://en.wikipedia.org/wiki/Router_(computing)
"A router is considered a Layer 3 device because its primary forwarding decision is based on the information in the Layer 3 IP packet, specifically the destination IP address. "

instant000

alxx wrote: »

But if a router is a layer 3 device doesn't that make the protocols and applications it uses also level/layer 3 ?

http://en.wikipedia.org/wiki/Router_(computing)
"A router is considered a Layer 3 device because its primary forwarding decision is based on the information in the Layer 3 IP packet, specifically the destination IP address. "

OSI is a model, not a requirement. The key thing you're supposed to take from it is that it is a modular structure, designed to aid developers, and also to help you understand how networking works.

A lot of apps were designed for the four layer TCP/IP model, and don't have discrete functions that easily separate to each of the 7 OSI layers (This is especially problematic on the upper layers)

With regards to layer 7 apps being degraded to layer 3 because they end up going across a router, then by that same measure, everything can be reduced to layer 1, as it goes across the interface/wire.

We're over-analyzing this.

Several of us have commented that it's only a "model" and not a strict requirement that things fall neatly into.

I'd like to think that IT peeps like myself work at Layer 9. (Regular users work at layer 8, LOL.)

alxx

instant000 wrote: »

OSI is a model, not a requirement. The key thing you're supposed to take from it is that it is a modular structure, designed to aid developers, and also to help you understand how networking works.

A lot of apps were designed for the four layer TCP/IP model, and don't have discrete functions that easily separate to each of the 7 OSI layers (This is especially problematic on the upper layers)

With regards to layer 7 apps being degraded to layer 3 because they end up going across a router, then by that same measure, everything can be reduced to layer 1, as it goes across the interface/wire.

We're over-analyzing this.

Several of us have commented that it's only a "model" and not a strict requirement that things fall neatly into.

I'd like to think that IT peeps like myself work at Layer 9. (Regular users work at layer 8, LOL.)

ah a sheep herder

I'm more of an embedded systems guy (embedded systems/network sensor systems/network programming) .

As more and more switches get layer 3 switching , could argue that we're converging towards the tcpip model