OSI vs. TCP/IP models question

N3tWrkNut

This may be a silly question but I get the OSI model and I get the TCP/IP model layers. So maybe I am over thinking this but is the OSI primarily for LAN and TCP/IP (which I know is internet) is basically for WAN? Am I right coming to that conclusion?

JDMurray

The TCP/IP model is the oldest of the two networking models. It reflects the opinion that networking is primarily controlled in the lower layers of the TCP/IP stack. Also, "dumb" hardware is not represented by a distinct layer.

One of the innovations of the later OSI model was that it provided a more granular view of network communications by divvying-up the TCP/IP model's giant "Application" layer into smaller layers that reflect what really goes on in there from a "software-centric" point of view. An additional layer was also added that separates "dumb" hardware from "smart" hardware (e.g., hardware with microcode or firmware). The OSI model therefore provides a more balance representation of the stages of network communication processing.

So neither model is preferred based on networking technology (WAN, MAN, LAN, CAN, PAN, etc.). Instead, a networking customer will choose the model that it believe best represents its view of the computer networking universe and insist on the use of that model. In my experience, most everyone prefers the OSI model, except for US government agencies and the DoD, which use the TCP/IP model. Networking vendors usually support both to appeal to customers using either model.

N3tWrkNut

Thank you....that makes sense!

CSCOnoob

JDMurray wrote: »

The TCP/IP model is the oldest of the two networking models. It reflects the opinion that networking is primarily controlled in the lower layers of the TCP/IP stack. Also, "dumb" hardware is not represented by a distinct layer.

One of the innovations of the later OSI model was that it provided a more granular view of network communications by divvying-up the TCP/IP model's giant "Application" layer into smaller layers that reflect what really goes on in there from a "software-centric" point of view. An additional layer was also added that separates "dumb" hardware from "smart" hardware (e.g., hardware with microcode or firmware). The OSI model therefore provides a more balance representation of the stages of network communication processing.

So neither model is preferred based on networking technology (WAN, MAN, LAN, CAN, PAN, etc.). Instead, a networking customer will choose the model that it believe best represents its view of the computer networking universe and insist on the use of that model. In my experience, most everyone prefers the OSI model, except for US government agencies and the DoD, which use the TCP/IP model. Networking vendors usually support both to appeal to customers using either model.

Hmmmmm. IIRC, all my networking books said OSI model was used to create the TCP/IP model. *shrugs*

JDMurray

CSCOnoob wrote: »

Hmmmmm. IIRC, all my networking books said OSI model was used to create the TCP/IP model. *shrugs*

The TCP/IP model was derived from the ARPA model, which was in use at least since 1973, and they are the same model. Work on the OSI model didn't start until 1976-77, which was specifically designed to be a more general representation of networking than the model ARPA had devised.

I'm researching when the TCP/IP model itself was formalized. I wouldn't be surprise if it was on January 1, 1983, when TCP/IP formally replace NCP as the standard protocol suite of the Internet. That would explain why people would infer that the TCP/IP model came after the OSI.

CSCOnoob

JDMurray wrote: »

The TCP/IP model was derived from the ARPA model, which was in use at least since 1973, and they are the same model. Work on the OSI model didn't start until 1976-77, which was specifically designed to be a more general representation of networking than the model ARPA had devised.

I'm researching when the TCP/IP model itself was formalized. I wouldn't be surprise if it was on January 1, 1983, when TCP/IP formally replace NCP as the standard protocol suite of the Internet. That would explain why people would infer that the TCP/IP model came after the OSI.

Cool. That'll be great!