Transcript NetProgCourse_ch1

Introduction
to Network Programming in
UNIX & LINUX
© Daniel Zinchin
Tel-Ran Ltd 2008-2009
© D.Zinchin [[email protected]]
Instead of Preamble…
‫בס"ד‬
Rabbi Avraham Yaakov 100 years ago taught his Chassids:
- It is possible to learn from anything. Everything in this world exists to edify us.
Not only what the Lord has made, but also what people have made, makes us wise.
- What, for example, do we learn from a railway? - one Chassid asked in doubt
- That, having been late for an instant, it is possible to miss entirely.
- And telegraph?
- That each word is taken into account.
- And phone?
- That everything told by you here, is audible there.
And what could we learn from the
INTERNET ?
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-2
Course Contents
•Network Primes
•Inter-Process
INTERNET
Independent Network
Communication (IPC)
•Inter-Host
Host
Process
Communication (IHC)
Thread
•Multi-Threading and
Synchronization
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-3
Recommended Literature and References:
1. W. Richard Stevens.
UNIX Network Programming,
Prentice Hall, 1990.
W. Richard Stevens
personal site:
www.kohala.com
2. W. Richard Stevens.
UNIX Network Programming, 2nd Edition,
Prentice Hall, 1998-99.
Vol.1 Networking APIs
Vol.2 Interprocess Communications
3. W. Richard Stevens.
TCP/IP Illustrated.
Addison-Wesley, 1994-1996
Vol.1 The Protocols
Vol.2 The Implementation
Vol.3 TCP for Transactions, HTTP, NNTP,
and the UNIX Domain Protocols
4. Douglas E. Cormer.
Internetworking with TCP/IP
© D.Zinchin [[email protected]]
With gratitude to my Tel-Ran teacher Kolman Shkolnik,
whose book was used during the preparation
of materials for this course.
Kolman Shkolnik.
Introduction to Internetworking in UNIX.
Tel-Ran 1995
Introduction to Network Programming in UNIX & LINUX
1-4
The History
<1960
1960
1965
1969
1971-75
1976
1978
1980
1983
1983
1984-86
1986
1988
End 80s
1988
1989
1991
1992
1994
2000
>2000
- Data transportation in the form of bits
- Transportation of data packets
- Bell Labs, General Electrics and MIT develop Multix
(Multiplexed Information and Computing Service)
- Ken Thompson and Dennis Ritchie in AT&T begin to develop
UNICS (UNiplexed Information and Computing Service)
- First releases of UNIX in AT&T
- First UNIX network application UUCP (UNIX-to-UNIX copy program) .
File transfer and electronic mail. AT&T Version 7 UNIX (1978)
- Berkley Version 7 UNIX (Eric Schmidt, Berkley University)
File transfer, e-mail, remote printing
- DARPA (Defense Advanced Research Projects Agency) ARPANET,
TCP/IP protocols development for Berkley Unix
- 4.2BSD UNIX system (Berkley Software Distribution) with Socket Interface
- Richard Stallman announces the GNU Project, an attempt to create a free operating system
- AT&T System V UNIX with Transport Layer Interface (TLI)
File transfer, e-mail, remote printing, remote command execution.
- NSFNET (National Science Foundation) – remote access to super-computer network.
- 4.3 BSD UNIX. Available source code of DARPA TCP/IP, Xerox NS protocols.
- Microsoft Xenix, System V for Intel 8086, 80286, 80386 processors
- IEEE (Institute of Electrical & Electronical Engineers) defines
POSIX – standard operation system interface.
- Unix System V Release 4.0 (SVR4). Merge of AT&T System V with SunOS (4.xBSD derivative)
Provides ANSI compliant C compiler.
- Linus Torvalds (Finland) introduces Linux – freeware OS for PC.
- Sun introduces Solaris based on SVR4.
- Red Hat Linux released.
- Sun releases Solaris 8 having big success
- Multiple releases and popularity growth of Linux derivatives
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-5
Basic Terms
Computer Network
Communication system for connecting end-systems. Enables to
share data, programs and resources (distributed systems).
There are physical networks and logical networks.
Host
Single computer, end-system. Could be personal computer, dedicated
system (print or file server) or time-sharing system.
Process
Any program which is executed by computer’s operation system.
Thread
Separate part of process, providing it’s specific working flow,
and sharing the process data and resources with other threads.
Inter-Process Communication
Sharing of information and resources by two or more different processes.
Inter-Host Communication
Communication between two or more processes, running on different
hosts in the network.
Communication Protocol
Set of rules and conventions that communication participants must
follow.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-6
What is Internet
Internet
•Set of interconnected independent computer networks.
•Uses common suite of protocols called TCP/IP.
•Managed by groups of representatives:
•IAB
-Internet Activity Board
•NIC
-National Information Center
•FNC
-Federal Network Center
RFC - Request For Comments
Internet Services
Transport Level:
•Unreliable packet delivery
•Reliable stream transport
Application Level:
•File Transfer (FTP, TFTP)
•Electronic Mail (SMTP)
•Remote Login (TELNET)
•Network File System (NFS)
•Remote Program Execution
•Shared peripheral devices
© D.Zinchin [[email protected]]
The name of the result and the
process for creating a standard on the
Internet.
The proposals are reviewed by the
Internet Engineering Task Force.
New standards are proposed and
published on the Internet, as a Request
For Comments with acronym RFC <#>.
Introduction to Network Programming in UNIX & LINUX
1-7
Layering and Protocol Family
Example of Layering in Communication System.
Layering is decomposition of task
into subsystems (pieces), designed
as sequence of horizontal layers.
As result, each layer:
• concentrate on providing a
particular function
• is built in terms of one layer
below
• provides means to building
various types of upper neighbor
layer.
Protocol Family (Suite) is set of
interfaces between layers or inside
layer.
Peer-to-Peer Protocol is the
protocol used between two entities of
the same layer.
© D.Zinchin [[email protected]]
Transfer of Knowledge by Means of
Inter-Person Verbal Communication
Knowledge
Knowledge
Concepts
Concepts
Speech
Speech
Voice
Voice
Transmitter
Receiver
Sound Waves
Introduction to Network Programming in UNIX & LINUX
1-8
7-Layer OSI Model
7
6
5
4
3
2
1
Application
messages
Presentation
messages
ISO
(kinds of compression)
Session
OSI Model
Open System Interconnection Model
messages
International Standard Organization
(dialog management)
Transport
datagrams
(inter-process level)
Network
packets
(inter-host level)
Data Link
frames
(network topology)
Physical
© D.Zinchin [[email protected]]
bits
The 7-Layer OSI Model, developed by ISO
(1984), is the guide, providing a detailed
standard for describing of a network .
Advantage of layering is to provide well-defined
interfaces between the layers, when change
in one layer doesn’t affect an adjacent layer.
Introduction to Network Programming in UNIX & LINUX
1-9
4-Layer Simplified Model of TCP/IP
TCP/IP Protocol Suite actually was developed (1980-83) before formulation of 7-layer
ISO OSI Model (1984).
It implements 4-layer Simplified Communication Model:
7
6
5
4
3
2
1
© D.Zinchin [[email protected]]
Application
Presentation
messages
Application
(communication
end-point)
(kinds of compression)
Session
4
(dialog management)
Transport
datagrams
(inter-process level)
(inter-process level)
Network
packets
frames
(network topology)
Physical
Network
(inter-host level)
(inter-host level)
Data Link
Transport
bits
3
2
Data Link
(network topology &
physical connection)
Introduction to Network Programming in UNIX & LINUX
1
1-10
Protocol Suite For 4-Layer TCP/IP Model
Application
Transport
application layer protocols
transport layer protocols
network layer protocols
Network
data link protocols
Data Link
Application
messages
Transport
datagrams
Network
packets
Data Link
frames
physical connection
bits
Protocol Suite for 4-Layer TCP/IP Model
contains 4 types of peer-to-peer protocols:
• Application Layer protocols –
• Transport Layer protocols –
• Network layer protocols –
• Data Link protocols –
© D.Zinchin [[email protected]]
end-point communication
inter-process communication
inter-host communication
topology-specific interface with physical network
Introduction to Network Programming in UNIX & LINUX
1-11
TCP/IP Model and Protocol Suite
TFTF
Trivial File Transfer Protocol
UDP
User Datagram Protocol
TCP
Application 1
user application
Application 0
TFTP
Provided by
kernel of OS
UNIX
Transmission Control Protocol
Internet Protocol
ICMP
Internet Control Message Protocol
Transport
ARP
UDP
IP
ICMP
Address Resolution Protocol
Network
Reverse Address Resolution
Protocol
Data Link
TCP
ARP
RARP
messages/
stream
datagrams
packets
IP Address + Port
IP
Ethernet
Token Ring
frames
Ethernet
A local area network architecture
with broadcast bus topology
physical connection
bits
Physical Address
RARP
Token Ring
A local area network architecture
with ring topology and token
passing scheme
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-12
Network
LAN
Local Area Network is a computer network
comprises a local area, like a home, office, or
group of buildings.
MAN
Network
Type
Wide Area Network is a computer network
covering a broad geographical area.
Largest and most well-known example of a
WAN is the Internet.
Speed
LAN
Coaxial cable,
fiber optics,
Wi-Fi (wireless
technology)
MAN
Coaxial cable,
microwave link
56 Kbit/s –
155 Mbit/s
WAN
Telephone lines,
microwave link,
satellite channels
9.6 Kbit/s –
45 Mbit/s
Metropolitan Area Network is large
computer network usually spanning a city.
WAN
Technology
4 Mbit/s –
2 Gbit/s
Gateway is a system, that
interconnects two or more
networks
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-13
Communication Activities
• Data Transmission
Communication Networks can be divided into two basic types by method of data
transmission: circuit-switched and packet-switched.
• Encapsulation
Encapsulation is hiding of object data from rest of the world. For protocol suite this
means adding of control information to data when going one layer down.
• Multiplexing and Demultiplexing
Multiplexing means “to combine many into one”. For network this means combining of
data accepted from different functionalities of neighbor layer.
Demultiplexing is reverse of multiplexing.
• Routing
Routing is making decision, what route the packet should take.
Static Routing is based on precomputed information.
Dynamic Routing is depends on state of network configuration in the specific moment of
time.
• Fragmentation and Reassembling
Fragmentation (or segmentation) is breaking up of a packet into smaller pieces
(MTU – maximal transmission unit)
Reassembling is reverse of fragmentation, it is restoring of original packet from smaller
pieces used for transmission.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-14
Data Transmission Method
Communication Networks can be divided into two basic types by method of data transmission: circuitswitched and packet-switched.
Circuit-Switched Data Transmission
circuit
• Non-shared dedicated communication line is established
• Information transmitted without division
• Connection established once, then all data transmitted through this connection.
Packet-Switched Data Transmission
hop
hop
hop
hop
hop
• Shared communication links are used instead of dedicated line
• Information is divided into pieces – packets.
• Each packet contains the address of destination and separately routed over shared data links.
The TCP/IP Internet uses packet-switched data transmission,
provided by IP (Network) layer, responsible for forwarding of IP packets.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-15
Data Encapsulation in TCP/IP
Encapsulation is hiding of object data from rest of the world.
For protocol suite this means adding of control information to data
when going one layer down.
Data
Application
data
Data
TFTP
message
UDP TFTP
header header
Data
UDP
datagram
IP
UDP TFTP
header header header
Data
IP
packet
Ethernet
IP
UDP TFTP
header header header header
Data
Ethernet
trailer
400
4
TFTP
header
Bytes: 22
20
© D.Zinchin [[email protected]]
8
4
Example. Data encapsulation
during mail delivery
address
Ethernet
frame
Introduction to Network Programming in UNIX & LINUX
1-16
Multiplexing and Demultiplexing in TCP/IP.
Example.
Multiplexing means “to combine many into one”.
For network this means combining of data accepted from different functionalities of neighbor layer.
Demultiplexing is reverse of multiplexing.
Multi-homed Multi-user Host
Process 1
Process 2
Process 3
Process 4
Process 5
Multiplexing
UDP
TCP
IP
TCP/IP Suite
XNS Suite
Ethernet
interface
Demultiplexing
Ethernet
interface
Ethernet cable 2
Ethernet cable 1
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-17
Routing
Router is “intelligent” gateway, making a decision, what route (path) the packet should take.
Remote packet
is sent to Router
Application
Local packet
is sent to Recipient
TCP
TCP
hop 1
IP
Data Link
IP
(router)
hop 2
IP
(router)
hop 3
IP
Data Link
Data Link
Data Link
LAN1
Host 1
Application
LAN2
Gateway 1
LAN3
Gateway 2
Host 2
In TCP/IP routing is made on IP Layer. Each packet could have its own route.
The TCP/IP Internet uses Distributed Dynamic Routing.
Distributed Dynamic Routing
uses a mixture of global and local information to make routing decision.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-18
Fragmentation and Reassembling
Fragmentation
is breaking up of a packet into smaller pieces (transmission units).
Maximal Transmission Unit (MTU)
is maximal packet size held by network layer, depending on Data Link characteristics
Reassembling
is reverse of fragmentation.
MTU=128 b
MTU=2 Kb
Packet
1 Kb
IP (router)
fragmentation
128
b
128
b
128
b
...
MTU=2 Kb
128
b
128
b
128
b
reassembling
Packet
1 Kb
Data Link
Data Link
LAN 1
IP (router)
WAN connection
Gateway 1
LAN 2
Gateway 2
In TCP/IP fragmentation and reassembling is done at IP layer.
The IP layer performs these activities depending on requirement of specific Data Link layers,
hiding the technological differences between the networks.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-19
Client-Server Model
Service
requests
provides
is
described
by
Client
uses
Server
uses
Protocol
Client-Server Model is standard model for network applications.
Typical Client scenario:
Typical Server scenario:
• Starts and connects to network
• Starts and connects to network
• Sends service request to known server
• Waits for request from potential clients
• Accepts the service
• Accepts requests and provides service
• Waits for the next potential client…
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-20
Modes of Communication Service
Communication Service provided between two peer entities at any layer of the OSI Model.
Connection Mode
Connectionless Service
•
Provides hop-by-hop transmission of separate
messages.
Datagram delivery service
•
Each message transmitted independently and
contains all the information (address) required for
delivery.
hop-by-hop
Connection-Oriented Service
•
•
Virtual circuit service
Provides establishment of dedicated end-to-end
end-to-end
virtual circuit for data transmission.
Connection-oriented data exchange involves three following steps:
o
Connection establishment (performed once, requires overhead activity)
o
Data transfer (can be lengthy)
o
Connection termination
The dedicated circuit is called virtual, because it could be provided even on network with packet-switched data
transmission.
A connection-oriented service is often used when more than one message is to be exchanged between the two
peer entities.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-21
Modes of Communication Service (continuation)
Data Stream Format
Message Service
•Provides record boundaries .
Byte Stream
•Does not provide record boundaries.
Full-duplex - connection allows data to be transferred in both directions in the same time.
Half-duplex - connection allows data to be transferred in both direction,
but only one side to transfer at a time
Simplex
- connection allows data to be transferred only in one direction
(one end can only transmit and the other end can only receive.)
Reliability
Service is Reliable if it provides Sequencing and Error Control.
Most of reliable services provide also Flow Control.
Sequencing
• Means that the data is received by the receiver in the same order as it is transmitted by the sender.
In a packet-switched network, it is possible for two consecutive packets to take different routes, and thus
arrive at their destination in a different order from the order in which they were sent.
Error Control
• Guarantees that error-free data is received at the destination.
There are two conditions that can generate errors:
- the data gets corrupted (modified during transmission),
- the data gets lost.
The network implementation has to provide for recovery from both these situations.
Flow control (pacing)
•Assures that the sender does not send data at a rate faster than the receiver can process the data.
If Flow Control is not provided, it is possible for the receiver to lose data because of lack of resources.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-22
Summary: Communication Activities and Service Modes
Communication Activities
Model
Layer
Application
Protocol
Data
Multiplexing/
Encapsulation
Routing
Transmission
Demultiplexing
application
protocols
(applicationdepended)
TCP
hide
Application
layer data
Transport
UDP
Network
IP
Packet hide
switched
Transport
(hop-by-hop) layer data
Data Link
Ethernet/
Token
Ring
frame
hide Network
transmission
layer data
in LAN
© D.Zinchin [[email protected]]
combine/split of
data from
different
Transport
(south)
protocols
combine/split of
data from
different
Application
(north)
processes
Communication Service Modes
Fragmentation/ Connection Data Stream
Reliability
Reassembling Mode
Format
(applicationdepended)
(application- (application- (applicationdepended) depended) depended)
brake up
(/recompose)
stream into
(from) IP
packets
full-duplex
connection- (bioriented
directional)
byte stream
reliable
(sequencing,
error control,
flow control)
connection- datagram
less
delivery
unreliable
combine/split of
brake up
data from
distributed (/recompose)
different
connection- packet
dynamic
packet into
Transport
less
delivery
routing
(/from)
(north) and
transition units
Data Link
(south)
protocols
unreliable
combine/split of
data from
different
Network (north)
protocols
Introduction to Network Programming in UNIX & LINUX
1-23
Communication Services
Provided by TCP/IP Protocol Suite
Network Layer
IP – Internet Protocol
Provides unreliable connectionless packet delivery service, containing:
• Routing,
• Fragmentation / Reassembling,
• Multiplexing / Demultiplexing.
Works with different Data Link protocols and topologies, hiding the
technological differences between the networks.
Transport Layer
UDP – User Datagram Protocol
Provides unreliable connectionless datagram delivery service.
TCP – Transmission Control Protocol
Provides reliable connection-oriented full-duplex byte stream service
over unreliable connectionless packet-switched IP Network.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-24
Data Link Connection and Topology.
Transceiver
Host
Host Interface
Transceiver
Coaxial cable (ether)
This is communication device capable of transmitting
and receiving of signals.
Performs analog - digital - analog translation.
Host Interface
It provides physical address associated with interface
hardware and filters incoming packets
Bus Topology
The main characteristic of this topology is that it is a
passive structure: when a node is down, the
network is not affected.
Ring Topology
This kind of topology is less efficient and reliable
but it is quite cheap. As soon as two lines are cut
the network no longer works.
Star Topology
This topology is quite efficient and cheap. Most
small local networks is built on this model by using
a central Hub that connects computers together.
A hub can imitate different network topology
configurations.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-25
Data Link Equipment
A to B
A to C
A to B
A to C
C
All frames are
transmitted to
both segments
Local frames
are filtered out
Repeater
A
C
B
Bridge
B
A
Repeater is a hardware component that transmits
Bridge is a hardware component that filters
frames from one wire and places them on another.
frames according to destination address.
Bridges are used to connect multiple segments.
Repeaters are a simple way to extend a LAN segment.
Hub is a common connection point for
devices in a network.
Hub can imitate a bus or a ring or could be
more sophisticated. In this case it called
Switch.
© D.Zinchin [[email protected]]
Buss
Imitation
Ring
Imitation
Hub
I
Hub
O
I
O
O
I
O
I
I
O
I
O
O
I
Introduction to Network Programming in UNIX & LINUX
O
I
1-26
Data Link Protocols: Ethernet
Ethernet is broadcast bus technology with best effort delivery.
Was developed in the beginning of 70-s by Xerox corporation. Following development
and standardizing was performed by DIX (Digital, Intel, Xerox) in 1979-1980. In 1983
it was approved as standard by IEEE .
Currently Fast Ethernet is one of most popular LAN technologies.
IEEE (Institute of Electrical and Electronics Engineers) - professional world-wide society for electronics
and electrical engineers)
Technology description:
• Each host interface has preset unique 48 bit physical address.
• Transceiver senses when ether is in use and detects collisions.
• When data is transmitted, all hosts connected to the bus can hear
the transmission.
• In case of collision both hosts wait for a random amount of time, before
sending the information again.
P
Bytes: 8
DA
6
occurs when two devices on a
network try to transmit
information at the same time.
– Preamble for synchronizing. The last
byte is SFD – Start of Frame Delimiter
Ethernet Frame
P
Collision
SA
6
L/T
2
Data
46-1500
DA
– Destination Address
CRC
SA
– Source Address
4
L/T
– (Length/Type) Length of Data and
optional ID of upper level protocol
CRC – Cyclic Redundancy Check sum
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-27
Ethernet CSMA/CD algorithm
Begin
Frame Sending
1. Prepare Frame
2. Trial Counter = 0
CSMA/CD
Carrier Sense Multiple
Access with Collision
Detection
Wait
Random Delay Time
Is other node
transmitting ?
NO
Calculate
Random Delay Time
IPG
NO
Inter-Packet Gap
YES
YES
Is IPG interval
Passed ?
Is Trial Counter
Greater then 16 ?
Is other node
transmission
finished ?
NO
YES
Delay
NO
Wait until
IPG is passed
YES
JAM
Trial Counter ++
Send the 32-bit
JAM
Recovery
After Collision
YES
Is Collision
detected ?
YES
© D.Zinchin [[email protected]]
Send the Next
bit of Frame
NO
Is Last bit
sent ?
Frame Sending Failure.
Too many collisions.
Waiting
Send the 1-st
bit of Frame
32 bit frame for collision
signaling
NO
Transmission
Frame is Sent
Successfully
Introduction to Network Programming in UNIX & LINUX
1-28
Data Link Protocols: Token Ring
Token Ring is deterministic technology with predictable delay.
Was developed in 1980 – 1985 by IBM. Approved as standard by IEEE.
Multiple
Access
Unit
Technology description:
• This is not continuous wire, consists of connections among host
interfaces, connecting to Ring by means of Multiple Access Units
(MAU). (No more than 8 hosts per MAU).
• Physical address is configurable by means of switches.
• Control frame named “Token” is passed from one host to another,
allowing to this host (and only this) to send the packet.
• To send the frame, host performs the following steps:
• waits for the arriving Token
In Token Ring collisions never
• converts it to data frame and copies to the next host in the Ring
occur.
• waits for the frame to return after delivery
This ensures good performance
• deletes the frame and sends out a new Token
of network under big loads (30%• Each moment of time no more than 1 host sends the data,
40%)
all other hosts copy the data by chain.
• Host interface could be in following modes:
• transmit mode – sending host
• copying mode – all other hosts in ring
• recovery mode – recovery in case of token loss
• When copying host recognizes its destination address, it cleans refuse bit.
• Sender, accepting the original frame, detects if frame was delivered, checking the refuse bit.
Fist connected to Ring host accepts the status of Active Monitor. It responsible for:
• Token creation and recovery
• Check frame delivery timeout
• Deletion of frames not deleted by other hosts.
• Notification of other hosts about its presence in Ring (sends “Active Monitor Present” frame)
In case of Active Monitor problem, other hosts compete to accept its status.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-29
Token Ring (continuation)
Token Ring “Token”
Token Ring Data Frame
SD
AC
FC
DA
SA
Bytes: 1
1
1
6
6
Data
0..4096
CRC
ED
FS
SD
AC
ED
1
1
1
1
1
4
SD
- Start Delimiter
AC
- Access Control - packet priority, type (token/data), active monitor bit
FC
- Frame Control
DA
- Destination Address
SA
- Source Address
CRC - Cyclic Redundancy Check sum
ED
- End Delimiter - contains 1 bit – Last Packet bit, Error flag bit
FS
- Frame Status - contains Parity bit (copy indicator),
Refuse bit (destination reached)
ETR – Early Token Release technology
After sending of frame, the same host generates new Token. As result, many sequential frames could
circulate in the Ring in the same time. But no more than one Token could present in each moment of time.
This technology improves the performance of Token Ring.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-30
Network Layer: Internet Protocol
Internet Protocol (IP)
Provides unreliable connectionless packet delivery service, containing:
• Routing,
• Fragmentation / Reassembling,
• Multiplexing / Demultiplexing.
Works with different Data Link layers, hiding the technological differences between the
networks.
Application
application
application
Transport
TCP/UDP
TCP/UDP
Network
IP
address
IP
(host)
Data Link
Physical
address
Ethernet
hop 1
IP
(router)
hop 2
Ether Token
net Ring
Ether Ether
net
net
LAN
© D.Zinchin [[email protected]]
IP
(router)
LAN
Introduction to Network Programming in UNIX & LINUX
IP
(host)
hop 3
TokenRing
LAN
1-31
Internet Protocol: IP(v4) Packet Structure
bits:
1-st byte
5-th byte
9-th byte
13-th byte
17-th byte
• Version – IP Version
• Header length – IP Header total length in 32bit words (max = 15*4=60 bytes)
• DSCP, ECN – type of service fields, used by upper protocols
• Total Length – total packet length (header + data).
• Identification, DF (don't fragment),
MF (more fragments), Fragment Offset – used for fragmentation and reassembly.
• TTL - time-to-live – maximal number of hops, set by the sender, decremented by each router.
• Protocol - upper layer protocol (1=ICMP, 2=IGMP, 6=TCP, 17=UDP).
• Header Checksum - calculated over just the IP header including any options.
• Source IP address (32 bit)
• Destination IP address (32-bit)
• Options (<=40 bytes) , used by upper protocols
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-32
Internet Address Formats
Internet Address (IP address) is mandatory unique logical address which must have every host in Internet.
IP Address (IPv4) is 32-bit number.
Decimal Dotted Notation is human-oriented representation of IP Address as sequence of 4 decimal numbers
separated by dot.
The IP Address has internal structure. There are 5 classes of IP Address:
Bits:
1
Class A
0
Bits:
1
Class B
Bits:
Class C
Bits:
Class D
Bits:
Class E
7
24
Network ID
1
14
1 0
1
1
1
16
Network ID
1
1
1
1
8
Network ID
1
1
Multicast address
1
27
1 1 1 1 0
Class
Host ID
28
1 1 1 0
1
Host ID
21
1 1 0
1
Host ID
(reserved experimental format)
Range
Networks Per Class
*
Hosts Per Network
2^7 – 2 = 126
2^24 – 2 = 16,777,214
128.0.0.0 to 191.255.255.255
2^14 – 2 = 16,382
2^16 – 2 = 65,534
C
192.0.0.0 to 223.255.255.255
2^21 – 2 = 2,097,150
2^8 – 2 = 254
D
224.0.0.0 to 239.255.255.255
N/A
N/A
E
240.0.0.0 to 247.255.255.255
N/A
N/A
A
0.0.0.0
B
© D.Zinchin [[email protected]]
to 127.255.255.255
Introduction to Network Programming in UNIX & LINUX
*
*Note:
The 2 types of bit
sequences:
• All Bits equal 1
• All Bits equal 0
are not used as
Network IDs and
Host IDs
1-33
Netmask
Gateways, to locate the network, need only Network ID part of IP Address and don’t need to know the location
of every host. This is important concept of routing.
To calculate Network ID from IP Address, Gateways use Netmask.
NETWORK_ID = IP_ADDRESS & NETMASK
Class
Network ID
Netmask
A
1 byte
255.
0.
0.
0
B
2 bytes
255. 255.
0.
0
C
3 bytes
255. 255. 255.
0
Example.
Calculation of Network ID for IP Address = 145.11.99.243 (Class B)
Network ID
class B
1 0
0x
Host ID
0 1 0 0 0 1 0 0 0 0 1 0 1 1 0 1 1 0 0 0 1 1 1 1 1 1 0 0 1 1
9
1
0
145.
IP Address
B
6
11.
3
F
99.
3
243
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0x
Netmask
of Class B
F
F
F
255.
F
0
255.
0
0
0.
0
0
1 0 0 1 0 0 0 1 0 0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0x
Network ID
9
1
145.
© D.Zinchin [[email protected]]
0
B
11.
0
0
0.
Introduction to Network Programming in UNIX & LINUX
0
0
0
1-34
Network ID and Subnet ID
IP Addresses are assigned by specific authority:
InterNIC- Internet Network Information Center.
The InterNIC assigns only Network IDs. The assignment of Host IDs is responsibility of local site system administrator.
IP addresses are often subnetted. Subnet adds additional level to the address hierarchy:
• Network ID (assigned to site)
• Subnet ID (chosen by site)
• Host ID (chosen by site)
All the hosts on a given subnet share a common Subnet Mask, and this mask specifies the boundary between
the subnet ID and the host ID. Bits of 1 in the subnet mask cover the network ID and subnet ID, and bits of 0
cover the host ID.
SUBNETWORK_ADDR
= IP_ADDRESS & SUBNET_MASK
Example.
Subnetting of Network with Class B address, using 8 bit Subnet ID.
Bits:
1 1
Class B
1 0
Network mask:
14
Network ID
1 1 1 1 1 1 1 1
1 1
Class B
1 0
Subnet mask:
0 0 0 0 0 0 0 0
255.
0.
14
255.
0 0 0 0 0 0 0 0
=0xFFFF0000
0
8
Network ID
1 1 1 1 1 1 1 1
Host ID
1 1 1 1 1 1 1 1
255.
Bits:
16
8
Subnet ID
Host ID
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0
255.
255.
=0xFFFFFF00
0
In the example above local gateway needs only 8 bits of Subnet ID for routing. Adding new host to existing
sub-network will not require any changes to the internal gateways.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-35
Types of Destination IP Address
There are three Destination Types of IP addresses:
• Unicast
(destined for a single host)
• Broadcast (destined for all hosts on a given network)
• Limited Broadcast
to all hosts inside local network
(Network ID = all 1, Host ID = all 1)
• Net-Directed Broadcast
to all hosts of other local network
(Network ID = netID, Host ID = all 1)
• Subnet-Directed Broadcast
to all hosts in specific sub-network
(Network ID = netID, Subnet ID = subnetId, Host ID = all 1)
• Multicast (destined for a set of hosts that belong to a
multicast group or sub-network).
multicast
group
Multicast Address is a special type of address that is
recognizable by multiple hosts joined a Multicast Group .
A Multicast Address is sometimes known as a Functional Address
or a Group Address.
Hosts that are interested in receiving data flowing to a particular
group must join the group using:
• Broadcasting and Multicasting
needs support by hardware
(Data Link layer).
• Broadcast and Multicast
addresses could not be used as
Source IP Address
IGMP - Internet Group Management Protocol .
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-36
Special Case IP Addresses
Loopback Addresses
By convention, the address 127.0.0.1 is assigned to the Loopback Interface.
Anything sent to this IP address loops around and becomes IP input without ever leaving the machine.
This address id often used when testing a client and server on the same host.
Any address on the network 127/8 can be assigned to the loopback interface, but 127.0.0.1 is and is often configured
automatically by the IP stack.
(This address is known as INADDR_LOOPBACK )
Unspecified Address
The address consisting of 32 zero bits is Unspecified Address.
It is only permitted to appear as the source address in packets sent by a node that is bootstrapping before
the node learns its IP address.
(This address is known as INADDR_ANY).
Private Addresses
Three address ranges are set aside for “Private Internets“.
These are the networks that do not connect directly to the public Internet.
Small sites use these private addresses and Network Address Translation (NAT) to a single public IP address
visible to the Internet.
NAT - Network Address Translation
Also known as Network Masquerading or
Number of
IP-masquerading is a technique in which
Class Range
addresses
the source and/or destination addresses
of IP packets are rewritten as they pass
A
10.0.0.0
to 10.255.255.255
16,777,216
through a router or firewall.
B
172.16.0.0 to 172.31.255.255
1,048,576
C
192.168.0.0 to 192.168.255.255
© D.Zinchin [[email protected]]
65,536
It is most commonly used to enable
multiple hosts on a private network to
access the Internet using a single public
IP address.
Introduction to Network Programming in UNIX & LINUX
1-37
Multihoming and Address Aliases
Multihomed Host
This is a host with multiple interfaces.
Each interface must have a unique IP address. (Loopback interface is not counted)
A router, by definition, is multihomed since it forwards packets from one interface to another one.
But, a multihomed host is not a router unless it forwards packets.
There are two types of Multihoming:
• Physical Multihoming
Host has multiple physical interfaces, each interface has its own IP address.
• Logical Multihoming
Newer hosts have the capability to assigning multiple IP addresses to the same physical interface.
Each additional IP address, after the first (primary), is called an Alias or Logical Interface.
Multihomed Network
This is a network that has multiple connections to the Internet.
For example, some sites have two connections to the Internet instead of one, providing a backup capability.
ifconfig
UNIX/LINUX utility for
configuring network
interface parameters
© D.Zinchin [[email protected]]
$ ifconfig -a
eth0 Link encap:Ethernet HWaddr 00:11:25:0C:DE:88
inet addr:145.9.228.95 Bcast:145.9.228.255 Mask:255.255.255.0
inet6 addr: fe80::211:25ff:fe0c:de88/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2295264 errors:0 dropped:0 overruns:0 frame:0
TX packets:783513 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:319104166 (304.3 MiB) TX bytes:75720636 (72.2 MiB)
Base address:0x2000 Memory:e8100000-e8120000
lo
Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:804848 errors:0 dropped:0 overruns:0 frame:0
TX packets:804848 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:49311882 (47.0 MiB) TX bytes:49311882 (47.0 MiB)
Introduction to Network Programming in UNIX & LINUX
1-38
Domain Name System
Domain Name System (DNS)
This is a distributed database that provides the mapping between IP addresses and hostnames.
Hostnames are more suitable for human use, than IP addresses.
The DNS is distributed because no single site on the Internet knows all the information.
Each site maintains its own database of information and runs a DNS Server program that other systems across the
Internet (clients) can query. The DNS provides the protocol that allows clients and servers to communicate with each
other.
Applications access the DNS through a Resolver. The resolver contacts one or more name servers to do the
mapping.
nslookup
UNIX utility for
DNS information
access
$nslookup gate88.mot.com
Server: abcde.mot.com
Address: 145.19.17.68
Name: gate88.mot.com
Address: 145.19.238.87
The DNS Name Space is Hierarchical Tree.
The InterNIC maintains the top-level domains:
• Generic Domains (com, edu, gov, int, mil, net, org)
• Country Domains (us, uk, il, ru, etc.)
InterNIC delegates responsibility to others for specific Zones. A Zone is a subtree of the DNS tree that is
administered separately. Many second-level domains then divide their zone into smaller zones.
To accept the DNS information, every Name Server must know how to contact the Root Name Servers.
The root server tells the requesting server to contact another server, and so on.
A fundamental property of the DNS is Caching.
Name Server caches accepted {IP Address ; Hostname} information for following reuse.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-39
Address Resolution
Address Resolution is translation between Network Layer logical address and Data Link physical address.
Host A
IP Address
(logical)
Ethernet Address
(physical)
IPA
PHA
Host B
Network
IP
Data Link
Ethernet
IP packet
Ethernet
frame
Network
IP
Data Link
Ethernet
IPB
PHB
physical connection
Address Resolution Problem
I want to send IP packet to another host with known IP Address. What is Physical Address of that host?
• Known: IPA, PHA, IPB.
• Unknown: PHB
Reverse Address Resolution Problem
I’m diskless workstation. What is my own IP address ?
• Known: PHA.
• Unknown: IPA.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-40
ARP
ARP – Address Resolution Protocol
ARP Request
Solves Address Resolution Problem.
( [ IPA , PHA] , [ IPB , ? ] )
Provides dynamic mapping from an IP Address
to the corresponding Physical Address on the
same physical network.
ARP Request is Ethernet Broadcast frame
(Destination Ph Address = 0xFFFFFFFFFFFF) is
sent to all hosts in physical network.
X
A
Y
B
Only the host, recognizing its own IP Address in
the Request, sends the ARP Reply, containing its
Physical address.
Z
ARP Reply
( [ IPA , PHA] , [ IPB , PHB ] )
ARP Cache
• Maintains the recent mappings from Internet
addresses to Physical addresses.
X
A
Y
B
Z
• Removes its entries after expiration time
Proxy ARP
Lets to Router to answer ARP requests,
addressed to another physical network,
substituting router’s physical address instead of
target foreign host address.
arp
$arp –a
UNIX utility for
ARP Cache
access
sun (140.252.13.33) at 8:0:20:3:f6:42
svr4 (140.252.13.34) at 0:0:c0:c2:9b:26
Than, accepting the frame, router forwards it
to the target foreign host.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-41
RARP Protocols
RARP Request
RARP – Reverse Address Resolution Protocol
( [ ? , PHA ] )
Solves Reverse Address Resolution Problem. Used by
diskless hosts during its initialization (bootstrap).
RARP Request is Ethernet Broadcast frame ,sent to all
hosts in physical network.
X
A
Y
Z
RARP
Server
Only the RARP Server, containing the required information,
sends RARP Reply, containing IP Address of requestor.
RARP Server serves single LAN. It could be multiple RARP
Servers in LAN. They handle Distributed Data Base of IP
Addresses.
RARP Servers provide delay mechanism to avoid
simultaneous response to requestor from multiple RARP
Servers in the same time.
( [ IPA , PHA ] )
X
A
Y
RARP Reply
RARP
Server
Z
BOOTP – Bootstrap Protocol
• Enables a diskless workstation to discover its own IP address, BOOTP Server IP address, and a file to be
loaded into memory to boot the machine.
• Needs manual pre-configuration of the host information.
• Could be routed and serve more than one LAN.
DHCP – Dynamic Host Configuration Protocol
• Allows dynamic allocation of network addresses and configurations to newly attached hosts.
• Allows recovery and reallocation of network addresses through a leasing mechanism.
• Does not require manual pre-configuration of the host information.
• Could be routed and serve more than one LAN.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-42
ICMP
ICMP - Internet Control Message Protocol.
ICMP handles Error and Control information messages
between routers and hosts.
These messages are normally generated by and
processed by the TCP/IP networking software.
ICMP Message
ICMP Header ICMP Data
IP Header Type
Code
CRC
Example of usage: ping and traceroute programs use
ICMP.
Type
Description
Query
*
0
echo reply (ping reply)
3
destination unreachable
4
quench (flow control)
5
redirect (use another router)
8
echo request (ping request)
9
router advertisement (reply to solicitation)
*
*
10
router solicitation (request for advertisement)
*
11
time exceeded (TTL=0)
12
parameter problem (bad IP header)
13
time stamp request (what time is it now)
14
timestamp reply (current time is…)
*
*
17
address mask request (give my subnet mask)
*
16
address mask reply (your subnet mask is …)
*
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
Error
*
*
*
*
*
1-43
IP Routing
IP Routing is distributed. Each hop of the specific packet is calculated separately.
Remote packet
is sent to
Router
application
Local packet
is sent to
Recipient
application
TCP/UDP
TCP/UDP
IP
(host)
IP
(router)
hop 1
Ether
net
Ethernet
hop 2
Ether
net
Ether
net
LAN
IP
(router)
IP
(host)
hop 3
Token
Ring
LAN
TokenRing
LAN
traceroute
hop1
hop3
G
G
hop2
© D.Zinchin [[email protected]]
UNIX utility, printing the route
which packets take to network host
istanbul% traceroute sanfrancisco
traceroute to sanfrancisco (172.29.64.39), 30 hops max, 40 byte packets
1 frbldg7c-86 (172.31.86.1) 1.516 ms 1.283 ms 1.362 ms
2 bldg1a-001 (172.31.1.211) 2.277 ms 1.773 ms 2.186 ms
3 bldg4-bldg1 (172.30.4.42) 1.978 ms 1.986 ms 13.996 ms
4 bldg6-bldg4 (172.30.4.49) 2.655 ms 3.042 ms 2.344 ms
5 ferbldg11a-001 (172.29.1.236) 2.636 ms 3.432 ms 3.830 ms
6 frbldg12b-153 (172.29.153.72) 3.452 ms 3.146 ms 2.962 ms
7 sanfrancisco (172.29.64.39) 3.430 ms 3.312 ms 3.451 ms
Introduction to Network Programming in UNIX & LINUX
1-44
IP Routing: Routing Table
Routing Table
The IP layer has a Routing Table in memory that it searches each time it receives a datagram to send.
The Routing Table contains the following information:
• Destination
– IP Address of Destination Host (flag “H”) or Destination Network (no flag ”H”)
• Gateway
– IP Address of Hop Router for Remote Network (flag “G”)
or IP Address of Local Host in Directly Connected Network (no flag “G”)
• Flags
– “U”- route is up, “G”- route to Remote Network via Gateway, “H” - route to specific Host,
“D”- route was added because of an ICMP Redirect Message.
• Ref
– The number of times the route used to establish a connection.
• Use
– The number of transmitted packages
• Interface
– The Network Interface used by route
UNIX Utilities:
route
- Utility for manual manipulation with Routing Table
netstat - Utility, showing network status
interface
Host
Gateway
Destination
# netstat -nr
Routing Table: IPv4
Destination
Gateway
Flags Ref Use
Interface
------------- ------------ ----- --- ------- --------127.0.0.1
127.0.0.1
UH
1
298
default
175.16.12.1
UG
2
50360
175.16.12.0
175.16.12.2
U
40
111379
175.16.2.0
175.16.12.3
UG
4
1179
- Route to Remote Network 175.16.2.0 via Gateway 175.16.12.3
175.16.1.0
175.16.12.3
UG
10
1113
- Route to Remote Network 175.16.1.0 via Gateway 175.16.12.3
175.16.3.0
175.16.12.3
UG
2
1379
- Route to Remote Network 175.16.3.0 via Gateway 175.16.12.3
175.10.4.3
175.16.12.5
UGH
1
1119
- Route to Remote Host 175.10.4.3 via Gateway 175.16.12.5
© D.Zinchin [[email protected]]
lo0
- Loopback Route (Local Host) via interface lo0
- Default Router 175.16.12.1
bge0
- Directly Connected Network 175.16.12.0 via interface bge0
Introduction to Network Programming in UNIX & LINUX
1-45
IP Routing Algorithm
Single IP Routing algorithm for Host an Router
Most multi-user systems today (almost every Unix system),
can be configured to act as a Router. This means, that Host
and Router could have the same routing algorithm.
Note: Unlike Router, the Host never forwards packets
from one of its interfaces to another.
Make Routing Decision
Is Destination IP Address Found ?
no
yes
Extract Network ID
Is Local Network ?
no
yes
Is Subnet Mask Specified ?
no
yes
Extract Sub-Network ID
Is Sub-Network ID Found ?
yes
no
Is Network ID Found ?
yes
Is Directly Connected Network ?
yes
no
Is Default Route Found ?
no
no
yes
Get Gateway IP Address
Find Directly Connected Network
Get Interface
Discard IP Packet.
Send ICMP Error.
Send IP Packet
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-46
Static and Dynamic IP Routing
Static Routing
During Static Routing the Routing Table:
• created during interface configuration
• added by the route command
• or created by an ICMP redirect (if the wrong “default” was used)
It is fine if the network is small, has a single connection point to other networks
and does not have redundant routes (which could be used if a primary route fails)
Dynamic Routing
During Dynamic Routing the Routing Table also updated by Routing Daemon process.
• the Routing Daemon is running on the Router
• communicates with another Routers using a Routing Protocol
• dynamically updates the kernel's routing table with information it receives from neighbor
routers
Routing Protocols are separated to:
• IGP – Interior (Intra-Domain) Gateway Protocols
Used between the Routers of one autonomous system.
Examples: RIP – Routing Information Protocol , OSPF - Open Shortest Path First protocol.
• EGP – Exterior (Inter-Domain) Gateway Protocols
Used between the Routers of different autonomous systems.
Example: BGP – Border Gateway Protocol
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-47
IP Routing Schema
IP Input
Routing
Daemon
Updates from
adjacent
Routers
route
netstat
command
command
IP Output
Routing Table write
Manual
updates
by Sysadmin
ICMP
Redirect
UDP
TCP
ICMP
IGMP
Routing Table read
Deliver Message
yes
yes
Is IP Forwarding
Enabled ?
no
no
Routing
Table
IP output: Calculate Next Hop
(to Destination Host or Hop Router)
Is Hop
Calculated ?
yes
Send IP Packet.
IP Layer
Is this My Packet ?
(Destined to my IP Address
or Broadcast Address)
no
yes
Process IP options.
Is this Source Routing
(prescribed by sender) ?
no
Discard IP Packet.
Send ICMP Error.
Receive IP Packet.
Put it to IP Input Queue
Physical Network Interfaces
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-48
Transport Layer: Port Numbers
Port Number is unique 16-bit identifier of Application
Communication Endpoint, used by Transport Protocols.
application
Application
TCP ports and UDP ports are independent.
Servers are normally known by their well-known port number.
For example:
• FTP server always uses TCP port 21.
• Telnet server is on TCP port 23.
• TFTP server is on UDP port 69.
Clients usually need any unique free port. Client port numbers are
called ephemeral ports.
Transport
Port
number
TCP/UDP
Network
IP
address
IP
Data Link
Physical
address
Ethernet/
TokenRing
Port number reservation:
•
1 –
1023 Well-Known Ports.
Used by well-known servers across all Internet.
• 1024 – 49151 Registered Ports.
Used by servers known across specific networks.
• 49152 – 65535 Dynamic Ports.
Used as ephemeral port numbers.
The ranges of Registered and Dynamic Ports are configurable.
File /etc/services
on most Unix systems
the well-known port
numbers are contained
in this file.
© D.Zinchin [[email protected]]
Transport Protocols provide
multiplexing / demultiplexing
based on Port Number
Port 1
#grep telnet /etc/services
telnet 23/tcp
#grep domain /etc/services
domain 53/udp
domain 53/tcp
Introduction to Network Programming in UNIX & LINUX
Port 2
Port 3
Transport Protocol
(UDP/TCP)
IP
1-49
Association and Transport Address
Communication link between two
processes is completely specified by
5-tuple Association:
• Transport Protocol (UDP / TCP)
• Local IP Address
• Local Port
• Foreign IP Address
UDP Association Example
Process A
Process B
application
application
UDP
Port
A
Port
B
UDP
IP
IP Addr
A
IP Addr
B
IP
• Foreign Port
Transport Address (or Socket)
is half-Association:
data link
data link
• Transport Protocol (UDP / TCP)
Socket A = { UDP, Port A, IP Address A }
• Local IP Address
Socket B = { UDP, Port B, IP Address B }
• Local Port
Association = { UDP, Port A, IP Address A, Port B , IP Address B }
Socket Pair defines the Association
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-50
Transport Layer: UDP Protocol
UDP – User Datagram Protocol
Provides unreliable connectionless datagram delivery service.
UDP is a simple protocol: each output operation by a process produces exactly one UDP
datagram, which causes one IP datagram to be sent.
UDP Datagram Structure
•Source and Destination Port Numbers - identify the sending process and the receiving process.
• UDP length
- the length of the UDP header and the UDP data in bytes.
(it is redundant, because could be calculated from IP header)
• UDP checksum
- covers the UDP Pseudo-Header (see below) and the UDP data.
To let UDP double-check that the data has arrived at the correct destination, the UDP checksum is calculated on
UDP Pseudo-Header, containing:
 UDP header itself
 IP Header fields: Source IP Address, Destination IP Address, Protocol type.
• DATA field may be empty
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-51
Transport Layer: TCP Protocol
TCP – Transmission Control Protocol
Provides reliable connection-oriented full-duplex byte stream service
over unreliable connectionless packet-switched IP Network
TCP provides reliability by doing the following:
• The application data is broken into TCP Segments - the best sized chunks passed by TCP to IP.
Each TCP Segment has its Sequence Number.
• When TCP receives data from the other end of the connection, it sends an Acknowledgment.
• When TCP sends a segment it maintains a Timer. If an acknowledgment isn't received in time, the segment is
retransmitted.
• TCP maintains a Checksum on its header and data. If a segment arrives with an invalid checksum, TCP discards
and doesn't acknowledge it, expecting the retransmission from sender after expiration of its timeout.
• TCP Re-Sequences the data if necessary, passing the received data in the correct order to the application.
• TCP provides discarding of duplicate data.
• TCP provides Flow Control. Each end of a TCP connection has a finite amount of buffer space. A receiving TCP
allows the sender to send as much data as the receiver has buffers for.
TCP Connection Establishing.
Three – Way Handshake.
active
side
Simple data exchange example.
TCP Connection Termination.
Modified Three – Way Handshake.
passive
side
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-52
TCP Segment Structure
• Source and Destination port numbers - identify the sending process and the receiving process.
• Sequence Number – unique identifier of TCP Segment, equal to the sequence number of segment 1st data byte in
stream between TCP sender and TCP receiver. Byte numeration begins from ISN (initial sequence number)
chosen by TCP sender. Wraps back around to 0 after reaching 2^32 - 1.
• Acknowledgement Number – used with ACK flag. The number of next byte, which TCP receiver is ready to receive.
• Header Length - the length of the header in 32-bit words.
• Flags: URG – urgent data (used with Urgent Pointer),
ACK – acknowledge (used with Acknowledge Number)
PSH - flush receiver data from its cache to process, RST – reset the connection (“port unreachable” reply)
SYN – connect (used with Sequence Number = ISN), FIN – finalize the connection
• Window Size – number of bytes, which receiver is ready to accept (flow control)
• TCP Checksum - covers TCP Pseudo-Header (TCP Header + IP Addresses and Protocol fields) + TCP Data
• Urgent Pointer – used with URG flag. Offset from Sequence Number to last byte of urgent data.
• Options (<=40 bytes). Maximal Segment Size (MSS) announcement, Timestamp, Window scale, etc.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-53
TCP Sliding Window
Sliding Window is algorithm of positive acknowledgement with re-transmission.
Receive Buffer
TCP Receiver
Receive Window
TCP Segments
1
ACKed,
accepted
by application
2
3
4
…
…
…
…
…
…
…
…
ACKed,
NOT accepted
by application
Sliding
direction
Send Buffer
TCP Sender
Send Window
TCP Segments
1
2
3
Sent,
ACKed
4
5
6
7
Sent,
NOT ACKed
8
9
10
Can Sent
immediately
11
…
Can’t Sent, until
window moves
• Receive Window – consists of any space in Receive Buffer that is not occupied by data.
Data remains in Receive buffer until target application will accept it.
The size of Receive Window is advertised by TCP Receiver to TCP Sender.
• Send Buffer
- begins from first un-acknowledged segment. Has the size of advertised Receive Window Size.
• Send Window
- covers unused part of the Send Buffer.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-54
TCP State Transition Diagram
The Legend:
normal transition for Client
normal transition for Server
appl:
state transition when application
issues operation
recv:
state transition taken when segment
received
send:
what is sent for this transaction
TIME_WAIT State
The endpoint that initiates termination of
connection, goes through TIME_WAIT
state and remains there for the period:
2 * MSL (maximum segment lifetime)
MSL is configurable life time of IP packet
in the network. The actual duration of
2MSL varies from 30 sec up to 4 min.
The TIME_WAIT state has two reasons:
1.Reliable termination of TCP connection
(ability to reply ACK for resent FIN)
2.To allow old duplicate segments to
expire in the network
(1MSL for original duplicate segments
+ 1 MSL for replies
to be lost from network)
For the period of 2MSL the TCP
prevents the incarnation (the restart
from the same port) of the connection.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-55
Modern protocols in TCP/IP Protocol Suite
IPv6 - Internet Protocol version 6.
Network Layer protocol, designed in the mid-1990s as a replacement for IPv4. The major change is a larger
address comprising 128 bits to deal with the explosive growth of the Internet in the 1990s.
IPv6 addresses are 128 bits long and are usually written as eight 16-bit hexadecimal numbers.
IPv6 provides packet delivery service for TCP, UDP, SCTP, and ICMPv6.
ICMPv6 - Internet Control Message Protocol version 6.
Is integral part of IPv6 implementation.
ICMPv6 combines the functionality of ICMPv4, IGMP, and ARP.
SCTP - Stream Control Transmission Protocol.
Transport Layer protocol, providing Reliable Connection-Oriented Full-Duplex Message Service.
Designed in 2000-2002.
SCTP connection called Association, because SCTP is Multihomed and involves a set of IP addresses and
a single port for each side of an Association.
SCTP provides a Message Service, which maintains record boundaries. As with TCP and UDP, SCTP can use
either IPv4 or IPv6, but it can also use both IPv4 and IPv6 simultaneously on the same association.
SCTP can provide multiple streams between connection endpoints, each with its own reliable sequenced
delivery of messages. A lost message in one of these streams does not block delivery of messages in any of
the other streams.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-56
Virtual Network and Tunneling.
The 6bone.
The 6bone is a virtual network that was created in 1996 for users with islands of IPv6-capable hosts wanted to
connect them together using a virtual network without waiting for all the intermediate routers to become IPv6capable. The 6bone is established on top of the existing IPv4 Internet using tunnels.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-57
TCP/IP Summary
IPv4
Internet Protocol version 4.
IPv4 Applications
IPv6
IPv6 Applications
Internet Protocol version 6.
tcpmdump routed
TCP
ping
traceroute
appl.
appl.
appl.
appl.
appl.
appl.
traceroute
ping
Transmission Control Protocol.
API
UDP
User Datagram Protocol.
TCP
SCTP
UDP
SCTP
Stream Control Transmission Protocol.
ICMP
ICMP
Internet Control Message Protocol
(version 4).
IPv4
IGMP
128-bit
address
32-bit
address
IPv6
ICMP
v6
v6
ARP
RARP
IGMP
Internet Group Management Protocol.
BPF
DLPI
ARP
datalink
Address Resolution Protocol.
RARP
Reverse Address Resolution Protocol.
ICMPv6
Internet Control Message Protocol version 6.
BPF
BSD Packet Filter. Provides access to the datalink layer in Berkeley-derived kernels.
DLPI
Datalink Provider Interface. Provides access to the datalink layer in System V R4.
© D.Zinchin [[email protected]]
Introduction to Network Programming in UNIX & LINUX
1-58