The Design Philosophy of The Darpa Internet Protocols

This paper describes the design philosophies for the Internet protocol suite, TCP/IP which was first proposed 15 years ago by DARPA. It shows us the design goals of the Internet with their importance and how these goals led to the evolution of Internet. Author motivated on the creation of the Internet architecture and how the protocol designs were selected and reasoning behind the design decisions. He discussed in-detail about the goals of the Internet.

The primary goal for them was to develop an effective technique for multiplexed utilization of existing interconnected networks. The initial networks to be connected used packet switching, packet switching was adopted as a fundamental component of the Internet architecture. A packet switched communication in which number of networks are connected using gateways which implements store and forward algorithm. This goal clearly says that ‘what’ should be achieved, i.e. “effective communication” but does not offer clear picture of characteristics of an effective system. Therefore, the second level goals were presented in the priority order:

1. Internet communication must continue despite loss of networks or gateways.
2. The Internet must support multiple types of communications service.
3. The Internet architecture must accommodate a variety of networks.
4. The Internet architecture must permit distributed management of its resources.
5. The Internet architecture must be cost effective.
6. The Internet architecture must permit host attachment with a low level of effort.
7. The resources used in the Internet architecture must be accountable.

The above goals were stated in the order of importance, and an entirely different network architecture would result if the order were changed. This is the point which I liked about this paper. It was interesting to know that if the goals or their ordering was different, the Internet would have been designed quite differently. The author gives a detailed description of top three goals. The first one among these is “Surviving in the face of failure” which meant that Internet should continue to supply communication service, even though networks and gateways are failing i.e. communication between two entities should never be lost. In the case of failure, to synchronize the communication without ending up the connection we must have a way to store the state information stored somewhere in the lower layers like intermediate nodes and replicate it to avoid loss of data. But the designers used a much simpler solution called as ‘fate-sharing’. Fate-sharing suggests gathering all the information and store it at the end points of the network at the entity which is utilizing the service of the network.

The second goal was “Handle multiple types of communications service”. The services here are like speed, latency, reliability, etc. For example, a remote login requires small delay but less bandwidth while file transfer was not more concerned of delay but of throughput. Basically, they both used the bi-directional reliable data delivery service (Virtual Circuit) which was the first service provided in the internet. Author also mentioned that there are few types of services that did not fit directly into the Internet like XNET (Cross-Internet debugger) and VOIP. This was when they knew that something more than TCP was needed. XNET didn’t seem to be a good fit because we don’t need an architecture that says it requires a reliable transport. In a case of VOIP, continuous delivery of packets is more important that reliability. Thus, it was decided to split TCP and IP into two layers each one providing different type of service. This paper gave a very significant information on how TCP/IP were designed and developed. This is the second thing I liked about this paper.

The third goal was “Accommodating a variety of networks” which showed that internet should operate with variety of networks. The simplest way to do this is to make the requirements for integrating a network as simple as possible. Author briefly discussed other goals in the paper. It is also stated in the paper that Internet fundamentally uses datagrams which is transported across the network. It says that datagram is a building block and not a service, upon which other services are made of. The reason to use datagram is because they eliminate the need for connection state within the intermediate nodes and they provide variety of types of services. Also, it assumes minimum assumption, thus enabling wide variety of networks to be incorporated.

Author concludes by saying that, the protocols are widely used in the commercial and military environment. At the same time, its success has made clear that priorities of the designer do not match the needs of actual users. More attention must be paid to accounting, resource management and operations of regions with separate administrations are needed. Also, there may be a better building block than the datagram for the next generation of architecture. Author could have improved the paper by giving more details about the impact of network layering on the network architecture and how the other layers have evolved. The point which I disliked about this paper was author did not consider that the priority list ordering is not constant and it can change in future according to needs.