Unit 1 Discussion

SUBJECT OF DISCUSSION:

DOD TCP/IP and OSI networking reference models

Q: How are they the same in what they are describing?

  • Both models describe ways for computer networks to inter-operate.
  • Each model starts its “work” from the point at which an application makes a request to send or retrieve/receive data over the network.
  • Both models contain a “stack” of “layers” that each provide some service in the act of transmitting or receiving data over the network.
  • Each model “bottoms out” at a layer that concerns itself with the physical components involved in network communications.

Q: Besides the different number of layers in each, how are they different?

  • Based upon what I’ve read, it appears that the DoD’s model makes some direct assumptions regarding what specific types of applications will run on “the Internet.”  In contrast, the OSI’s model allows for complete freedom in what type of applications can communicate over the network, and offers some (comparatively) wide-ranging flexibility in how the varying supported (and supporting) protocols can operate.
  • What I’ve read on the DOD model also seems to indicate that their model didn’t really make any direct allowances for routing hardware to be “standardized” in their system.  While this is just my preliminary determination from “skimming” a page or two, OSI’s model pretty much directly makes some allowances for routing hardware standardization.
  • Because of the lack of application-independence, it could be argued that the DoD’s model could be much more reliable (less chance for things to go wrong in a much simpler system).
  • Similarly, it could be argued that the complexity of the OSI’s system could make it less reliable (or more expensive to produce solutions around).

Q: Are they really describing the same thing? If so, why do we have both? If not, what’s the difference?

To my way of thinking, both models describe the same thing.  I believe the reason we have both is because the DoD’s model seems to be much simpler, possibly reflecting the needs of “earlier days” of “the Internet.”  Nowadays, most network classes focus quite a bit (and primarily) on the OSI model.  My belief as to why the OSI model is our main “point of study” in these classes is due to the greater capability for network stability and ease of design that might be possible under the OSI’s more-clearly-segmented series of layers.  At the same time, however, I believe we must familiarize ourselves with both of these models due to the relevance of the DoD’s version in how we have arrived at the levels of capability of the inter-networking systems we see today.

Q: How do they each relate to networking computers?

Both of these models try to set standards for how differing computer systems can communicate with one another over a network-specifically a Wide Area Network (WAN).  They provide methods for structured communications setup, reliable transmission, reliable reception, and communications tear-down.  A chief “selling point” of both of these models are that they allow resilient (fault-tolerant) networks to be (relatively) easily built using products from different manufactures.  And, finally, both of these models make available (and/or possible) a suite of well-known services many of us still use today (30 or 40 years later).

REFERENCES:

Pages referred to in the writing of this Discussion post include:

“TCP/IP Architecture and the TCP/IP Model”. The TCP/IP Guide. 02 March 2010 <http://www.tcpipguide.com/free/t_TCPIPArchitectureandtheTCPIPModel.htm>.

“DoD Networking Model”. Connected: An Internet Encyclopedia. 02 March 2010 <http://www.freesoft.org/CIE/Course/Section1/5.htm>.