why teach the OSI model?

Toggle sidebar Toggle sidebar
mammador

mammador

Platinum Member
Dec 9, 2010
2,120
1
76
I've been looking over CCNA and CCDP syllabus material, and they still teach the OSI model. But why? DoD, TCP/IP, Internet Protocol Suite (whatever one wants to call it) are the protocols actually in use. Why learn seven layers when 4 layers do the same thing o_O;)
 
D

drebo

Diamond Member
Feb 24, 2006
7,034
1
81
Cause telling people "It's a Layer 8 problem" is funnier than telling them "It's a Layer 5 problem".

Seriously, though, while the TCP/IP stack model is accurate, the OSI model is also still accurate and better termed.
 
E

Emulex

Diamond Member
Jan 28, 2001
9,759
1
71
they should teach objection relational model now :) network topology is so old school
 
C

cmetz

Platinum Member
Nov 13, 2001
2,296
0
0
The protocol suite that maps best to the OSI model is OSI.

Nobody uses it. Total marketplace failure, even with massive government assistance pushing for it.

(ok, yeah, I know, IS-IS is actually used inside many ISPs, but that's a historical accident)
 
D

drebo

Diamond Member
Feb 24, 2006
7,034
1
81
There are lots of L4 protocols that are not TCP.

GRE, for instance, is used all over the place. IPSec is built upon several non-TCP L4 protocols. Anything that runs on UDP is not TCP.
 
alkemyst

alkemyst

No Lifer
Feb 13, 2001
83,769
19
81
mammador said:
I've been looking over CCNA and CCDP syllabus material, and they still teach the OSI model. But why? DoD, TCP/IP, Internet Protocol Suite (whatever one wants to call it) are the protocols actually in use. Why learn seven layers when 4 layers do the same thing o_O;)
Click to expand...

They show you the other stacks too. Cisco speaks OSI.

You will be dealing with pretty much layers 1-4 at that level.

If OSI is your stumbling block, it's not going to get easier past it.
 
S

spidey07

No Lifer
Aug 4, 2000
65,469
5
76
There are MANY protocols out there that use layers 5 and 6 as well as 7. The OSI model is the very heart and foundation of networking and it must be slammed into one's head until you know it like your mother. Look at HTTPS.
 
G

Genx87

Lifer
Apr 8, 2002
41,091
513
126
Guess I never knew there was a push to get rid of it. I have always referenced the OSI model and so have my coworkers. Cant think of a single time anybody used the TCP\IP model when discussing layers.

For me, OSI makes more sense as it is more granular.

/shrug
 
T

theevilsharpie

Platinum Member
Nov 2, 2009
2,322
14
81
Genx87 said:
Guess I never knew there was a push to get rid of it. I have always referenced the OSI model and so have my coworkers. Cant think of a single time anybody used the TCP\IP model when discussing layers.

For me, OSI makes more sense as it is more granular.

/shrug
Click to expand...

The problem with the OSI model is that it's a model for protocols that never achieved wide-spread acceptance. Many network engineers (particularly Cisco-trained engineers) reference the OSI model because that's what they've been taught, but the engineers that actually develop the protocols used on the Internet don't care about it, and will use the TCP/IP model, if they use a model at all.

That's why the OSI model seems like a logical fit until you get up to the Layer 5 (where TCP/IP and OSI diverge), where it gets a bit awkward to classify different protocols. Don't believe me? When was the last time you had to resolve a layer 6 issue? Without Googling, Wikipedia, or looking at your notes, can you even tell me what protocols fit into layer 6? Yeah, that's what I thought :p

Some application stacks might be programmed with the OSI model in mind, and its components might logically fit into the OSI model, but that is nowhere near universal. However, any network running TCP/IP will fit into the TCP/IP model just fine.
 
alkemyst

alkemyst

No Lifer
Feb 13, 2001
83,769
19
81
In the real world most reference OSI that manage networks, it helps when you speak in the same terms.
 
S

spidey07

No Lifer
Aug 4, 2000
65,469
5
76
theevilsharpie said:
The problem with the OSI model is that it's a model for protocols that never achieved wide-spread acceptance. Many network engineers (particularly Cisco-trained engineers) reference the OSI model because that's what they've been taught, but the engineers that actually develop the protocols used on the Internet don't care about it, and will use the TCP/IP model, if they use a model at all.

That's why the OSI model seems like a logical fit until you get up to the Layer 5 (where TCP/IP and OSI diverge), where it gets a bit awkward to classify different protocols. Don't believe me? When was the last time you had to resolve a layer 6 issue? Without Googling, Wikipedia, or looking at your notes, can you even tell me what protocols fit into layer 6? Yeah, that's what I thought :p

Some application stacks might be programmed with the OSI model in mind, and its components might logically fit into the OSI model, but that is nowhere near universal. However, any network running TCP/IP will fit into the TCP/IP model just fine.
Click to expand...

Netbios/SMB
 
G

Genx87

Lifer
Apr 8, 2002
41,091
513
126
theevilsharpie said:
The problem with the OSI model is that it's a model for protocols that never achieved wide-spread acceptance. Many network engineers (particularly Cisco-trained engineers) reference the OSI model because that's what they've been taught, but the engineers that actually develop the protocols used on the Internet don't care about it, and will use the TCP/IP model, if they use a model at all.

That's why the OSI model seems like a logical fit until you get up to the Layer 5 (where TCP/IP and OSI diverge), where it gets a bit awkward to classify different protocols. Don't believe me? When was the last time you had to resolve a layer 6 issue? Without Googling, Wikipedia, or looking at your notes, can you even tell me what protocols fit into layer 6? Yeah, that's what I thought :p

Some application stacks might be programmed with the OSI model in mind, and its components might logically fit into the OSI model, but that is nowhere near universal. However, any network running TCP/IP will fit into the TCP/IP model just fine.
Click to expand...

That is actually pretty funny. I was talking to my boss about this after reading this thread. And we were discussing that exact same thing. I primarily deal with layers 1-5 and admit above that it becomes a quagmire. But at the same time dont remember anybody discussing the TCP\IP model over the OSI model.

And yes, at one point I was Crisco trained. But dont expect me to be able to do much on them anymore. It has been years. But apparently the OSI model stuck in my head :D
 
K

kevnich2

Platinum Member
Apr 10, 2004
2,465
8
76
I'm not by any means an expert on the OSI model but for me, it helps with troubleshooting. This I got directly from spidey and took it to heart - always work from layer 1 of the OSI model and work your way up. If you find you have a problem at layer 1 or layer 2, no sense in troubleshooting layer 3 until you have the issue at 1 or 2 fixed.

It's definitely helped me in understanding how the packets get from point A to point B and if something's not flowing correctly, reverse it and troubleshoot from layer 1 one up. Never really referenced the TCP/IP model yet.
 
mammador

mammador

Platinum Member
Dec 9, 2010
2,120
1
76
alkemyst said:
They show you the other stacks too. Cisco speaks OSI.

You will be dealing with pretty much layers 1-4 at that level.

If OSI is your stumbling block, it's not going to get easier past it.
Click to expand...

Oh, I know the model of the bat. but i don't get why Cisco still teaches it, when it's not in use. The DoD model sends data the exact way, so why teach OSI? o_O

If somebody sends an e-mail via OSI, the data flow process is the same in TCP/IP. Application layer (HTTP, SIP) is Application/Presentation/Session in OSI. Transport layer is the same in both, and the Internet layer in DoD is Network in OSI.
 
Gryz

Gryz

Golden Member
Aug 28, 2010
1,551
204
106
cmetz said:
(ok, yeah, I know, IS-IS is actually used inside many ISPs, but that's a historical accident)
Click to expand...

And because IS-IS rocks the socks off of OSPF and EIGRP.

It has cleaner concepts.
It has a simpler spec (once you learn the terminology).
It is easier to implement. So developers can focus on what's important (robustness, scalability) in stead of worrying whether they get every crappy detail of OSPF done right.
It is (and was) easier to extend. MPLS traffic engineering extensions, IPv6, all got into the basic IS-IS. While OSPF had to come with OSPFv3.
It is easier to troubleshoot.
It has less completely useless features.

Heck. I even believe that the concept of having one address per box, in stead of one address per interface is a lot better. Although that is technically outside IS-IS.

IS-IS forever.
 
Gryz

Gryz

Golden Member
Aug 28, 2010
1,551
204
106
theevilsharpie said:
Many network engineers (particularly Cisco-trained engineers) reference the OSI model because that's what they've been taught, but the engineers that actually develop the protocols used on the Internet don't care about it, and will use the TCP/IP model, if they use a model at all.
Click to expand...

The engineers that have developed most protocols either work for cisco, have worked for cisco, or will work for cisco in the future.

The first 4 layers of the OSI model and the TCP/IP model are very similar (if not the same). The difference is in layers 5, 6 and 7. In TCP/IP it is harder to show where those layers are. But the functionality is there.

Layer 5 functionality seems to be part of TCP. So TCP is not just layer 4, but also layer 5.

Sun's Remote Procedure Calls has some functionality that looks like layer 6. Trying to make application data look like it's the same across different platforms. You might not be aware of it, but not all machines use the same byte-order. In fact, Intel-CPU based systems use the "wrong" byte order. There's something called "network byte order". Every protocol designer, from layer 2 to the applications, need to take this problem into account. You as a user don't worry about it. But for developers it must be a nightmare. Having a formal layer that takes care of this (like layer 6 in the OSI layer, or Sun's RPC libraries) could be very convenient.

And layer 7 does exist in the TCP stack. See it as SMTP, FTP, HTTP, etc. I would put Netbios/SMB also as at layer 7. And NFS. And many more.

So the TCP/IP stack does have more than 4 layers. I can see 6 out of the 7 layers in the OSI model come back in TCP/IP.
 
Gryz

Gryz

Golden Member
Aug 28, 2010
1,551
204
106
kevnich2 said:
always work from layer 1 of the OSI model and work your way up. If you find you have a problem at layer 1 or layer 2, no sense in troubleshooting layer 3 until you have the issue at 1 or 2 fixed.
Click to expand...

Funny, I always used to work exactly the other way around.

Does the application work ?
If not, do other applications work ?
If not, does ping to the remote host work ? (Testing layer 3).
If not, does ping to the next-hop/def-gateway work ? (Testing layer 2).
If not, does ARP work ? (Testing layer 2 kinda).
If there's no ARP, is the interface up ? Line protocol up ? Etc.
If not, then I'll start looking at cables and connectors.

Heck, just being able to look at cables last is worth working in this order. :)
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom