A Bit About Packet and Circuit Switching
Packet Switching. In packet switching networks, a connection is not established for the entire transmission. The transmission is broken up into packets that can each take a different path and then be reassembled at the destination. One of the major advantages of this is that links can carry multiplexed streams of a much larger variety of packets and keep their utilizations much higher.
One hopes, in fact requires, that the capacity of network links is much greater than what is required by only a single user's connection. If the capacity of the link is any greater than this you don't want to tie up this capacity by having it focused on a single transmission. This way, transmissions from different sources can make use of the same links rather than having links dedicated to end-to-end transmissions for their duration.
In some telecommunications networks there are reasons to have what is called circuit switching where such dedication of resources may be implemented. POTS - Plain Old Telephone Service requires switched circuits. When you place a telephone call only one switched path/circuit from your telephone to the one you are dialing and back is maintained for your exclusive use for the duration of your call. While the links can be used within different circuits at different points in time, when you are using it, it is dedicated to your phone call. To do this with the information contained in computer related transmission would be very wasteful of resources.
Now days in order to improve quality of service for certain sorts of computer transactions, the importance of having options that lie someplace in between packet and circuit switching are important. We will get into these issues later in the course.
Utilizations and Idle Time. Think of differences in the flow of information between
|Phone Conversation||Web Interaction|
|must be able to talk at the same time||not as important that both ends be able to "talk" at the same time|
|information must flow in steady stream to be recognizable||information can arrive in bunches and be reassembled to display page|
|conversations are generally continuous with very few breaks||user usually takes time to examine a page before initiating another interaction|
|little or no idle time||much idle time for network traffic, particularly after page has been displayed|
|This implies that we don't need to have a
dedicated connection across/within the network for a particular
web interaction. But we do need a dedicated connection
across/within the network for a phone conversation.
More will be said in class.
Some Implications. The implications of being able to work with packets for web and internet interactions are many. But, in general, it implies that many people can share the resources of the Internet. So much of the initial cost viability of the Internet was based on users being able to make use of excess capacity in the pre-existing POTS - Plain Old Telephone System.
Most everyone has experience with using a dedicated dial-up connection for the last mile. They also have experienced the frustration of having such a slow connection when they want to download and upload. But they may have not really noticed how often the connection is actually sitting idle as they do things such as the following
We will get into these sorts of issues much more throughout the semester.
But, now I want to get back to some of the major implications of using packets for Internet interactions.
Developing understanding at this level is terribly important since it has implications about the following among many other things.
These lists can go on and on and they are something we will address and readdress throughout the semester.