Introduction to Computer

Networks
Chen Yu
Indiana University

Basic Building Blocks for Computer

Networks
• Nodes
– PC, server, special-purpose hardware,
sensors
– Switches
• Links:
– Twisted pair, coaxial cable, optical fiber,
phone line, wireless radio channels
• Network: two or most hosts connected by
links/switches

Addressing and Routing

• Address: something that identifies a node
often a unique byte string.
• Routing: find a route (path) to the
destination node based on its address.
• Types of traffic/addressing
– Unicast: to a single destination node.
– Broadcast: to all nodes on the network.
– Multicast: to some subnet of nodes on the
network.

1
 Network Protocol
• A network protocol describes how
communication entities (sender/receiver) should
interact to accomplish a networking task.
– Format, order of messages sent and received
– Actions taken on message transmission, receipt
• Protocol define interface, not implementations
Allow two sides of a communication task to be
independently designed / implemented.

An example
• A human protocol • A computer protocol
hi
Person A Person B Build a connection?
computer A
hi computer B
response
Got the time? Get http://www.iu.edu

3:30
<file>

Network Layers
• Separation of network functions /features
allow independent design and
implementation.
• Layering is a vertical separation of network
functions / features.
Hidden the complexity: each layer
interface hides complexity in this layer and
layers below.

2
 Internet Architecture
Bottom-up:
• Physical: electromagnetic signals on the
wire.
• Link: data transfer between neighboring
network elements. E.g. Ethernet
encoding, framing, error correction
• Network: host-to-host connectivity. E.g. IP
routing and addressing
• Transport: host-to-host data transport
reliable data transport, congestion control,
flow control. TCP/UDP
• Application: anything you want to do on
computer networks. HTTP, FTP, SMTP

Protocols for the Internet

Architecture

• Difference between layers, protocols, protocol implementations.

Socket Programming

• Socket: the interface between application processes and underlying

networking systems.

3
 Socket Programming
• Socket: an OS interface into which
application processes can communicate
with remote application process.
• Two types of transport services via socket
APIs:
– Connection-oriented byte stream (TCP)
– Connectionless datagram (UDP)

Port Numbers
• Multiple sockets might exist in each host
• A port number identifies each such socket
in each host
• Usually, port numbers ranging from 0 to
1023 are called well-known port numbers
and are restricted.

TCP socket
• At the server:
server must have created socket (door) that welcomes
client’s contact
server process must first be running
• Client contacts server by:
Creating client-local TCP socket
Specifying IP address, port number of the server socket
When client creates sockets: client TCP establishes connection to
server TCP.
• When contacted by client:
server TCP creates a new connection socket for
communicating with the client, which allows server to talk
with multiple clients.

4
 Berkeley Sockets
• Operations:
fd = socket(): create the socket
bind(fd,port): binds socket to local port.
connect(fd,port): establish a connection to
a remote port
send(), recv(), write(), read(): operations
for sending and receiving data
close(fd)

Association
• The 5-tuple that completely specifies the
two processes that make us a connection.
{protocol, local-addr, local-process,
foreign-addr, foreign-process}

Server-Client Model

5
 UDP Socket
• UDP: connectionless
no handshaking
sender explicitly
attaches IP address
and port of destination
to each datagram.
• Datagram-oriented

TCP Segment Structure

Protocols vs. Implementation

• They specify:
– Syntax and semantics of messages exchanged.
– Rules for when and how processes send & respond
to messages.
• They don’t specify the implementation:
Programming languages, data structures
• The goal:
components implemented independently can
inter-operate with each other as long as they
follow the protocol specification.

6
 Network Applications and
Application-layer Protocols
• Network applications:
– running in end systems (hosts)
– distributed, communicating using network.
– use communication services provided by
lower layer protocols (TCP, UDP).
• Application-Layer Protocols
define interface between application
entities.

Network Applications
• Some commonplace applications
– Web and HTTP
– FTP, telnet and ssh
– E-mail: SMTP, POP3, IMAP
– DNS
– Distributed file sharing

HTTP and web

• HTTP: hypertext transfer protocol
• Web: the application using HTTP protocol
• Client/Server model
– Client: browser that requests, receives, and
displays web objects
– Server: web server sends objects in response
to requests

7
 HTTP and Web
• HTTP specifies
– Types of messages exchanged, e.g. request &
response messages
– Syntax of message types: what fields in messages &
how fields are delineated.
– Semantics of the fields, ie. Meaning of information
– Rules for when and how processes send & respond
to messages.
• Web client and server can inter-operate as long
as they follow HTTP
– Web client (brower): IE, Firefox
– Web server: Apache, MS IIS

Electronic Mail
• Two types of entities:
– Mail servers
mailbox contains incoming messages for
users
message queue of outgoing (to be sent) mail
messages
– User agents
compose, edit, read mail messages.
pine, outlook…

Electronic Mail
• Two types of protocols:
– Mail transfer protocol
from sender agent to the receiver’s mail
server (SMTP simple mail transfer protocol)
– Mail access protocol
the receiver pulls mails from server to agent.
E.g. POP3, IMAP, HTTP

8
 Example
• You use your PC to compose a message with
Bob’s address.
• Your PC sends the message to her mail server
through SMTP; the message is placed in a
message queue.
• Your mail server sends message to Bob’s mail
server through SMTP.
• Bob’s mail server places the message in Bob’s
mailbox
• Bob invokes his user agent to read the
message.

SMTP interaction between mail

servers
• S: 220 www.example.com ESMTP Postfix
• C: HELO mydomain.com
• S: 250 Hello mydomain.com
• C: MAIL FROM:<sender@mydomain.com>
• S: 250 Ok
• C: RCPT TO:<friend@example.com>
• S: 250 Ok
• C: DATA
• S: 354 End data with <CR><LF>.<CR><LF>
• C: Subject: test message
• C: From: sender@mydomain.com
• C: To: friend@example.com
• C:
• C: Hello,
• C: This is a test.
• C: Goodbye.
• C: .
• S: 250 Ok: queued as 12345
• C: QUIT
• S: 221 Bye

Example
• telnet servername 25
• See 220 reply from the server
• Enter hello, mail from, rcpt to, data

This allows you

• Send email without using a normal email
client
• …

9
 Mail Access Protocols
• SMTP: delivery/storage to the receiver’s
server
• Mail access protocol: retrieve from server
– POP: Post Office Protocol
– IMAP: Internet Mail Access Protocol
– HTTP: Hotmail, yahoo!mail

DNS: Domain Name System

• People: multiple identifiers
SSN – unique, for tax reporting
Name – Human friendly, easy to remember
• Internet hosts:
– IP address –used for addressing, routing on the Internet
– “name”: e.g. heart.psyc.indiana.edu – human friendly.
• Question: Map between IP addresses and names?
DNS query: find the IP address for a given name
core function for Internet applications
“ssh avidd-b.indiana.edu” vs. “ssh 192.76.168.10”
http://www.cnn.com vs http://64.236.24.20

DNS: decentralized
• Distributed database – implemented with
collaboration of many name servers
distributed all over the network.
No server has all name-to-IP address
mappings.
• Why not centralize DNS?
• What if massively replicating it?

10
 Query
• Recursive Query:
puts the burden of name resolution on the
contacted name server.

• Iterative Query:
contacted server replies with name of server to
contact
“I don’t know this name, but ask this server”

DNS Caching
• Once a name is learned by the name server, it
caches mapping so the next query for the same
name can be answered directly.
This can happen at any step of the name lookup.
• Cache entries timeout (disappear) after some
time (e.g. two days).
Timeout is necessary because the mapping can
change.

DNS Scalability & Reliability

• Scalability
Poor scalability at few root name servers

• Reliability
Problems at a root server can cause big
troubles.
too many replicated root name servers
would make it difficult to synchronize
them.

11
 Distributed File sharing
• Napster: central index

• Gnutella (query flooding)

non-hierarchical, equal status

Peer-to-Peer Networks
• Fundamental advantage of p2p networks
better scalability – no performance
bottleneck.
better robustness – more tolerant to
random failures, intentional attaches.
• Challenge: peer coordination without
complete global knowledge.

Peer
• A pure peer-to-peer network does not
have the notion of clients or servers, but
only equal peer node that simultaneously
function as both "clients" and "servers" to
the other nodes on the network.

12
 Keyword search
• Users input a few keywords, the system returns a list of
documents matching the keywords. E.g. google
• Google maintains a central search index:
a search index contains a list of all searchable words,
each of which contains a list of documents relevant to
the word.
interaction document lists for multiple-word queries.
Scalability?
Robustness?

Peer-to-Peer Keyword Search

• Splitting the central search index into
smaller pieces and distributing them in
peer-to-peer fashion.
• Challenge: make them work
collaboratively to achieve similar speed
and quality of search.

Two solutions
• Splitting based on keywords
split the index database to many pieces
based on keywords and distribute them to
many nodes in the network.
• Split based on documents

13
 Putting them together
• Split based on keywords
weakness: transferring large data
segments for multiple keyword queries.
• Split based on documents
weakness: too many sites to visit for each
query.

14