OSI: The Internet That Wasn’t
I had fun this week talking with the students about the birth of the Internet, from its roots in the ARPANET through the multitude of networks that sprung up in the 1970s and 1980s. During our discussion, we gave a passing nod to Open Systems Interconnection (OSI) standard that was often compared to TCP/IP. OSI is a poster child for a standard by committee, and in this case the committee was hosted by the International Organization for Standardization (ISO). I was working at Honeywell in the mid-80s, a company supporting the ISO OSI standard, and I was taking evening courses toward my masters degree where I was learning about this “important” standard.
How does OSI and TCP/IP compare? There are lots of articles out on the net about this, but one of my favorite pictures is the following, found at http://fiberbit.com.tw/tcpip-model-vs-osi-model/:
As you can see, TCP/IP approaches the world in four pieces. First is the work done above TCP/IP, by an application that uses the network in performing its function. Below TCP/IP is another simple, single box labeled the “Network Interface” that is exactly what it says: the interface to the physical network. The middle is split between TCP (the Transport layer) and IP (the Network layer). On the OSI side, we have three layers above and two layers below the similarly represented (and well defined) Transport and Network layers.
I honestly never understood (and still don’t understand) why it was important to think about the single Application layer in the TCP/IP model as separate Session, Presentation, and Application layers in ISO. I’m sure that it was my own failing, but this was my poster child for unhelpful complexity. Sure, someone smarter than me can imagine applications where a Presentation and Session layer might be useful, but they couldn’t convince me that splitting out this functionality was universally important in the same way that we split the Transport and Network layers. Ok, maybe doing this splitting doesn’t need to be universally important for every application, but it would be nice if splitting out these layers was generative of new ideas.
Anyway, the purpose of this blog post is to encourage my students to read an IEEE Spectrum article from 2013 by Andrew L. Russell, whose title I lifted for the title of this blog post. You can find the article here. I hope you enjoy it as much as I did (and still do).