1.1 what is the Internet?

inter + network

ex. mobile network, home networks, institutional networks

component

H/W

end hosts: running network apps on terminals
interconnection devices: router, switch, repeater...
- packet switches: the way to communicate between hosts in the other networks
links: copper, fiber, radio, wireless...
- transmission rate: bandwidth ⬆ internet speed ⬆

S/W

Operating Software
application programs
protocols: rules for communications
- ex. message format, order, action

hosts => network edges => network core

1.2 Network Edge

1.2.1 Access Networks

: make network edges(⊃ end hosts) accessible to the network

architecture

my network	network core	my network
[end host(wireless) --(wireless)--- AP --(wire)-- edge router]	(routers)	[edge router -----end host]

transmission rate: bandwidth(bps) ⬆ internet speed ⬆
(central office) shared / dedicated

Digital Subscriber Line

dedicated access network

DSL modem -> splitter
DSL phone line
(central office) DSL Access Multiplexer
voice(~~4kHz): telephone network | data(4kHz~~): Internet Service Provider

Cable Network

shared access network

cable modem -> splitter
coaxial cable: cable company provide one signal to multiple users
(central office) Cable Modem Termination System
Hybrid Fiber Coax
Internet Service Provider

Home Network

cable or DSL modem, ONT -> splitter
Fiber To The Home
(central office) Headend
Internet Service Provider

Enterprise / Institutional Network

AP, Ethernet switch
Institutional link to ISP

Wireless Access Networks

shared access network

wireless LANs: Wireless Fidelity(shared network)
wide-area wireless access(shared network)

1.2.2 Physical Media

: the physical materials that are used to store or transmit information in data communications

guided: wired
- Twisted-pair cable: copper wires, behind the ethernet
- Coaxial cable: copper conductor, broadband
- Fiber-optic Cable: high speed, low error rate
unguided: wireless
- radio link types

1.3 Network Core

connecting network edges(⊃ end hosts)
the mesh of interconnected routers
when sending packets, forwarding from a router to the next router
- routing: Get a path from source host to destination host using routing algorithms
- forwarding: Move packets from the before router to the next router using forwarding table

1.3.1 Two fundamental approaches to moving data

comparison

circuit switching	packet switching
call setup, resource reservation : the resources needed along a path	No call setup, no resource reservation
multiple users sharing a link by Frequency Division Multiplexing , Time Division Multiplexing	full link capacity by dividing into small packets which have their own each destination address using store-and-forward transmission
low speed	3.5x users but after it.. low Quality Of Service
in telephone networks	~

store-and-forward transmission

app in source Host ---link(by packets)---> app in destination Host
how? storing bits until a packet and then forwarding it
=> total time

1.3.2 A Network of Networks

Internet = A Network (by ISP) of Networks (end system)

hierarchy structure

host < access ISP < regional ISP (peering link) < Internet eXchange Point < Tier 1 ISP or Content Provider

content provider has its own content provider network in its own data center

1.4 Evaluation Metrics in networks

1.4.1 delay

: source~destination seconds

nodal processing delay: before forwarding a packet, should check its error and destination address
- opt to the quality of router
queueing delay: when forwarding one packet, the other packet should wait in a buffer.
- opt to the number of network users => so variable
- traffic intensity = La/R ↑ ~1 infinite
  when packet arrival rate(packet/sec) > output link capacity(L/R)
  <=> over the queue -> dropped
transmission to the link delay when starting to forward: A host send a packet of L bits over a link with transmission rate R bits/sec, then the time to transmit the packet is L/R sec. (link transmission rate = link capacity = link bandwidth)
- opt to bandwidth of output link
propagation delay in the link: time = length / speed
- opt to the amount of data
=> end-end delay: the number of hop * L/R

1.4.2 loss

: Packet Loss Rate (<-> Packet Delivery Rate)

usually in router
in a buffer when queueing delay, data > queue capacity => dropped
host: re-transmission, network: waste of resource, user: delay

1.4.3 throughput

: sum of traffic <=> how much data has been moved between server and client
=> rate: bits/time unit

instantaneous: throughput at the peak
average: throughput on the average
- end-end average throughput == bottleneck link == min(R_s, R_c, R/10)

1.5 protocol layers, service models

Internet protocol stack

5 layers	explanation	protocol	encapsulation	controlled by
application	support network application	HTTP, SMTP, DNS(#24-DNS), FTP	message	user(app developer)
transport	data transfer process <-> process	TCP, UDP	+ segment	OS
network	find path	IP, routing protocols	+ datagram	OS
link	data transfer by hop from source to destination	Ethernet, WiFi	+ frame	OS
physical	on the wire like cable, radio	bits	Protocol Data Unit	OS

layering: modularization -> maintenance, system update
encapsulation: message + headers..
- source -> switch(link+physical) -> router(network+link+physical) -> destination(only message + segment are from source)
- always start from physical layer and go up by figuring out the needed info by the headers

2.1 principles of network applications

network apps(ex. gmail, youtube, zoom, game) work only on end systems

App Structure

model	communication	data consumer	data provider	scaling
client-server model	process of client <-> process of server	client on/off dynamic IP address	server always on permanent IP address	data centers↑
Peer-2-Peer model	arbitrary end system <-> arbitrary end system	all devices	all devices	peer↑(self-scalability)

Socket

A network socket is a software structure within a network node of a computer network that serves as an endpoint for sending and receiving data across the network.
door: application -> (socket) -> transport -> ...

Addressing Processes

for http message to web server

host device -> (permanent 32bit IP address + port number(process identifier)) -> host device

well known port numbers: 80(HTTP), 25(mail)

Protocol

email: SMTP > TCP
remote terminal access: Telnet > TCP
web: HTTP > TCP
file transfer: FTP > TCP

application layer protocol

open protocols: SMTP, HTTP, FTP, Telnet
proprietary protocols: skype

App-layer protocol defines

type: request/response message
syntax: fields of message
semantics: how to interpret fields
rules

underlying transport protocol

Transport Service Requirements

Transport Service	kinds of app	for
data integrity	file transfer, web transactions	no loss (<-> loss tolerant)
timing	Internet telephony, interactive games	time sensitive (<-> delay)
throughput	multimedia	minimum throughput (<-> elastic)

Internet transport protocols services

Transport layer protocols	+	-	occasion
TCP	reliable	complex -> low speed	email, web, file transfer
UDP	fast speed	unreliable	one-time transaction : streaming multimedia, internet telephony

TCP service: connection-oriented by handshaking
- error control: ~until no data error
- flow control: sender considers receiver's data capability
- congestion control: no data overload in router/switch

2.2 Web and HTTP

words

word

HyperText Transfer Protocol
Round Trip Time
Uniform Resource Locator
Carriage Return, Line Feed

HTTP response time

www: Webpage > base html file(frame) > objects > url(obj file loc)
HTTP layers: application: HTTP > transport: TCP > network: IP > link: ethernet, wifi

non-persistent HTTP

first object
1. 1RTT for tcp connection request/response: new TCP connection socket created to initiate tcp connection
2. 1RTT for http request(URL)/response(base HTML file: file transmission time): that socket deleted to terminate tcp connection
second object: 2RTT + file transmission time

-> 4RTT+α => long latency
-> then... how about parallel TCP connection: 4 sockets(< OS) => overhead

persistent HTTP

: no tcp connection more after initiated it

first object
1. 1RTT for tcp: two tcp connection sockets created
2. 1RTT for http: keep two sockets ()
second object
- 1RTT for http: parallel objects request/response

-> 3RTT+α

HTTP format

request

request line

: 1 line

method + URL(after hostname) + version + cr + lf
- ex. GET /100sun/1.md HTTP/1.0\r\n

header lines

: 1<= line

field name + : + value + cr + lf
- ex. Host: www.github.com\r\n
- ex. user-agent, accept-language, keep-alive(how long)
cr + lf (end of header lines)
- ex. \r\n

body

: optional - like POST

entire body

cookies

client can keep user session state in cookie file ex. my id in ebay, amazon

HTTP request
- if first access: A server creates ID
- if not: send my id by each server with cookie headerline
HTTP response msg

caching

web caches = proxy server

client for origin server
server for client

why?

overhead of origin server in

HTTP request/response

client <-> proxy server <-> origin server
if same request: client <-> proxy server

delay

need to be fixed

assumptions

data rate from origin servers to browsers
- object size * request rate
- 0.1 Mbits * 15/s = 1.50Mbps
access link rate
- 1.45Mbps

consequences

LAN utilization
- data rate / LAN availability
- 1.5Mbps / 1Gbps = 0.15%
access link utilization
- access link rate / date rate
- 1.5Mbps / 1.54Mbps = 0.99%
total delay
- RTT from institutional router to origin servers
  - 2s
- access delay
  - access link rate / date rate
  - 1.5Mbps / 1.54Mbps ≈ 1 → ∞ = minutes
- LAN delay
  - μs

Files

SN_chp2.md

Latest commit

History