Why We Build Networks
The Important
Throughout the millennia, reducing latency has increased economic activity. This was most evident in the formation of cities that shortened distances between a large population of people.
We transmit information using electro-magnetic waves because they travel through space much faster than humans, and human developed transportation systems. As such, they reduce transaction times, and increase productivity.
We build (routed/switched) networks because point-to-point links from every person to every other person does not scale economically, technologically, or aesthetically.
In achieving something that is economically and technologically possible, guaranteed capacity and delivery is replaced with probabilistic capacity and delivery. Even in TDM switching networks, capacity is probabilistic.
In order to restore guaranteed capacity and delivery, additional information/complexity is added to networks.
Guaranteed capacity is not provided for every person who uses a network.
Network design aspires to ensure all information is delivered to the right destination, within a “reasonable” period of time
The N-squared minus one problem
Imagine for a moment, there were no networks.
You want to email Trudy, your friend. So you go to your smartphone, type out an email, and hit “send”, but the email goes no where. Why? Your smartphone is not connected to anything. Sending emails to Trudy is really important. So you set up a radio transmitter at your place, and you install a radio receiver at Trudy’s place, so now you can send Trudy and email. Of course, Trudy will want to respond, so you also install a transmitter at Trudy’s place, and a receiver at your place.
Bob, another friend, hears about this email thing. He wants in. Ok, you say, I will install another transmitter and receiver at my place and at Bob’s place. Why another transmitter and receiver? Because there are no networks, therefore all communication is point to point, between every sender and receiver.
This email thing has been going fairly well for a few months, you have a hundred transmitters and receivers at your house, and another hundred people have transmitters and receivers at their houses. You are really saving a great deal of time with this email thing, but the cost of transmitters and receivers is mounting, not to mention the electricity bill. You are also a little worried about the radiation.
You are so impressed with this email thing that you decide to tell everyone in your 50,001 person town that they should email you when they want to buy something. You don’t know which of the 50,000 other people in the town may want to email you in the future, so just in case any one of them wants to, you install a transmitter and receiver at each of their houses, and 50,000 transmitters and receivers at your house.
The people in the town have come to really enjoy this email thing, but they are frustrated you are the only one they can use email with, so the town decides email is a basic human right, and installs 50,000 transmitters and receivers at each house. Of course, every time someone in the town wants to send an email to someone new, they have to disconnect their laptop from one pair of transmitters and receivers, and connect the laptop to another pair of transmitters and receivers. Some time after all 50,000 transmitters and receivers had been installed, a new application is created, called streaming video. This requires faster communication. So now all 50,000 transmitters and receivers at each house in the town have to be replaced so they can send and receive faster.
Now extrapolate this way of communicating to 5,6, or 7 billion people. This approach to communication would be both economical and technically infeasible, and that’s why we build networks.
Figure 1. Telephone wire mess in NYC as a result of telephone connections.
As most networking professionals know, this is not a theoretical problem, most have seen the pictures of New York City following the patent of the telephone, and before telephone switches were deployed. Point-to-point communication does not scale. Even with today’s networks, we are still eager to upgrade our networks to the most bits per second possible, so we can stop the number of cables getting out of hand, among other drivers.
We build networks because point-to-point does not scale economically, technologically, or aesthetically.
What Are The Tradeoffs?
If you build a point to point link to another person, you have control over how much capacity is available between source and destination, no one else can use the capacity, and it is always there when you want to use it. Arguably, you have better security/privacy. You at least do not have to worry about your communication being routed to the wrong destination, it is always going to the right destination, unless of course someone physically intercepts the communication.
In achieving something that is economically and technologically feasible, guaranteed capacity and delivery is replaced with probabilistic capacity and delivery.
In order to restore guaranteed capacity and delivery, additional information/complexity is added to networks.
Guaranteed capacity is not provided for every person who uses a network.
Network design aspires to ensure all information is delivered to the right destination, within an expected period of time
The very act of building a network with routers and switches is a decision to make tradeoffs. It is also a decision to achieve much more thant the
How do you take a theoretical network of billions of point to pint links and replace it with a network where each person only has to have a few or less links to the Internet?
Connecting networks user ports/cables with an information switch (Ethernet switch, IP router, TDM/Optical ADM/MUX,…)
Layers of aggregation
Increasing capacity in each layer (mostly)
The ability to determine how to get information from any source to any destination
An assumption that not every one will want to use the network at the same time
On the rare occasion that everyone does want to use the network at the same time, it frequently causes degraded performance: Major disasters, the first time American Idol told their audience to vote on their favorite singer, some past disasters, Mother’s day (historical, may not be still true), and similar events.
The very act of building a network with routers and switches is a decision to make tradeoffs. It is is also a decision to achieve a level of scale and utility that would not otherwise be possible.
In building networks we do make tradeoffs and lose some of the benefits of point to point links between family, friends, and business colleagues. However, we have gained much more than we have lost, especially in networks that are designed to restore some of what we have lost.