Saturday, August 15, 2015

Pancake of the Day: The Design Philosophy of the DARPA Internet Protocols

    If the paper of Cerf and Kahn focuses on the "now" (during their time) of the internet , David Clark's paper gives emphasis on the "tomorrow". He refers to the internet as "Internet" (with the big "i"), which I think means that during that time, it is already established that there is a global interconnection of networks. Also, if the motivation for the former's paper is connecting the exiting networks, this paper aims to explore the possibility of having a multi-media network.

   Since this paper was written circa 15 years after the birth of internet, it depicts some improvements from the original work. It's like the Chiffon cake that evolved from the flat wheat cake eaten by our ancestors before. 

   First, the author mentioned the fundamental goal for the Internet - "develop an effective technique for multiplexed utilization of exiting interconnected networks." He gave the metrics to define what an "effective technique" is.

   Second, he explained the goals in detail. For the objective of survivability, the author gave birth to the term "fate-sharing". It suggests that if one network lost the message to be communicated to the other network, there would be no need to keep the state information. This is the opposite of replication where in state information must be protected. He also highlighted the "datagrams" which are stateless packet switches in connection to "fate-sharing".

   The next goal explained was the support to different types of services. Here he explained that the Internet, to achieve this goal, should not reply only on TCP, but also in UDP (User Datagram Protocol).

   Moreover, another objective is not only to connect the existing networks but also other military and commercial facilities.

   Detailed descriptions of other miscellaneous objectives were stated in the paper.

   This is just my personal view on the paper. I found it somehow aggressive and questioning the original design for the internet. The author used the words "should be like this" or "should be like that", "would have been like this",  which I think gave the negative tone for me.

   On the other hand, I like how he saw the "future" of the internet. The goals that he enumerated were in-lined to that.
   Some questions that I had while reading the paper:

   1. On the discussion about survivability, the author said "the entities communicating should be able to continue without having to reestablish or reset the high level state of their conversation." What is a high level state of a communication?

   2. On the topic about fate sharing, it is said that "take this information and gather it at the endpoint of the net, at the entity which is utilizing the service of the network." Where is the endpoint? In the host? Where in the host? In the packet or segment? Will it be included in the ES or EM or REL in the header?

   3. On the explanation of TCP, it is indicated "this function was moved to the IP layer when IP was split from TCP, and IP was forced to invent a different method of fragmentation." I am curious to the new method of fragmentation that IP implemented.

   I ran out of flour :(

No comments:

Post a Comment