因为立即要申请出国了，所以在写这个系列的博客时大部分使用英文。
尽管是英文。但绝大部分内容都是我自己的感受和理解，供大家一起学习和讨论】
1 Network Edge
The device such as computers and mobiles connect to the Internet. So they are referred as end systems(who run the application programs) sitting at the edge of the Internet. And we use host and end system interchangeably, that is host=end system which is
divided into two categories: clients and servers.
2 Network Core
Network Core is the set of router-like and the Internet that implement the network and communication. And there are two type of Network Cores: circuit-switching and packet-switching.
2.1 Circuit Switching
Circuit-switching reserves dedicated circuit for each call or communication. When there are several circuits the network will divide the link bandwidth or switch capacity into several pieces - FDM and TDM.
In fact, the resource are divided according to the amount of users or links. For details in the example below: the 24 denotes that there are 24 circuits.
2.2 Packet Switching
Users share network resources in packet-switching. And the properties of packet-switching as follows:
- each packet uses full link bandwidth;
- Share network resources;
The free way of switching must take some disadvantages, as follows:
- the demanded resources in network may exceed the available - because of the second character
- the free link may cause congestions, build a waiting queue - because the packets are put into network sharing the resources and links.
The freedom is not absolute but relative. The resources in network are used on demand. So we also use the idea of multiplexing like circuit-switching. But the difference is that each slot or bandwidth is used on demand not fixed for a particular link or
something else.
Be care of the delay affairs in the Packet-Switching Network. Briefly, transmission delay -> propagation delay.
2.3 Packet Switching versus Circuit Switching
Packet Switching allows more users to use network.
The disadvantages of packet switching.
3 Delay, Loss, and Throughput
3.1 Four sources of packet delay
1. Nodal processing
In this part of work, the router or switch is going to process the packet to be sent. It will check the bit errors and determine the output link. So the processing cause some time consuming.
2. queueing
It is the time waiting at some output link to be transmitted. The length depends on the congestion level of each router.
La/R (a is the average packet arrival rate). The more the large time to wait.
a is about input, L/R is about output. Queueing delay is effected by both of input and output.
3. Transmission delay
We should distinguish the transmission and propagation first. The transmission is the work before propagation to transmit a packet to the next link.
There usually is a defination that: R=link bandwidth(1)(bps), L=packet length(bits), the time is L/R.
4. Propagation delay
The propagation delay is actually the time during which the packet will be propagated in the network.
There usually is a defination that: d=length of physical link, s=propagation speed in medium, the time isd/s.
NOTE: in some calculation, we simply assume that propagation is totally after the transmission. But this is just fit in low-speed links(which means the propagation take relatively long time), so we have the following formula
3.2 Packet Loss
This problem exists because the buffers of processing link has infinite capacity. Packets arriving the full buffered link will be dropped.
3.3 Throughput
We first distinguish the concepts of bandwidth and throughput.
Throughput is the rate at which the bits transferred between sender and receiver. So the throughput is a abstract concept that is just the result of a real Internet.
The throughput of a link is decided by the bottleneck link in it.
4 Protocol Layers and Service Models
4.1 Why layering
1. Build relationships between the complex components of a network.
2. Modularization.
============================================================================================
(1) bandwidth: 在我查看了维基并结合网络上带宽的使用，我认为在这里带宽的解释能够是这种：单位时间内，能够通过某个port的数据。