0% found this document useful (0 votes)
49 views30 pagesTransport Layer
Uploaded by
Dev KhatanharCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
49 views30 pagesTransport Layer
Uploaded by
Dev KhatanharCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 30
Computer Networks (Comp-MU)
4, Transit dela
Itis the time duration between a message being sent by the
transport user from the source machine and it being received
by the transport user atthe destination machine.
Residual error ratio:
— _Temeasures the number of lost or garbled messages asa
percentage ofthe total messages sent
= Weally the value of this ratio should be zero and
ractially it should be as small as posible,
6 Protection :
This parameter provides a way to protect the transmitted data
against reading or modifying it by some unauthorised partis.
1. Priority :
Using this parameter the user can show that some of ts
‘connections are more important (have higher priority)
than the other ones.
= Thisis important when congestions take place. Because
the higher priority connections should get service
before the low priority connections.
Resilience:
Due 10 internal problem or congestion the transport layer
spontaneously terminates a connection. The resilience
parameter gives the probability of such a termination,
6.4
Transport Service Primitives
“What are transport service primitives ? Discuss
ret. (Dec. 15, 10 Marks)
What are transport service primitives ? Explain.
ee (Dec. 17, 10 Marks)
= The transport service primitives allow the transport user such
2s aplication programs to access the transport service
= Bach transport service has its own acces primitives
= The transport service is similar to network service but there
are some important diferences. The main difference is that
the connection-oriented transport service is reiable,
~ The second difference between the network service and
transport service is whom the services are intended for. The
transport primitives are seen by many programs and
programmers. Hence the transport service is convenient and
easy to use.
~ We can get the idea about the transpor services by referring
to Table 64.1 which ists the five primitives,
Transport Lay
‘Table 6.41 + Primitives fora simple transport service
Sr] Priniive] TPDU sent Meaning
No. y
1, | LISTEN ‘None Block until some
process tres 0
@ [CONNECT | Connection | Actively attempt to
request establish a connection
3_| SEND Data Send data
RECEIVE | None Block until a data
[TPDU arrives
5. | DISCONNECT | Disconnection | Release the connection
request
| 0.2 What are transport service primitives ? Explain.
= The transport interface allows the application programs t0
establish, use and release connections.
= Let us see bow these primitives are used in actual
applications
1, The server executes a LISTEN primitive. This will
call to block the server until a client
2. When a client wants to tlk to the server it executes
the CONNECT primitive.
3. In response the transport entity blocks the caller and
sends a packet to the server. The transport laver
‘message is encapsulated in the payload of this packet
for the server's transport entity
Tepu:
‘The message sent from transport entity
called as transport protocol data unit or TPDU.
6.4.1 Nesting of TPDUs, Packets and
Frames : Pani
@,1. What are transport service primitives ? Discuss in
bret. (Dec. 15, 10 Marks)
to transport entity is
(Dec. 17, 10 Marks)
~The TPDUs which are exchanged by the transport layer are
conuinod in the pockets that ace exchanged by the network
layer.
~ These packets are in tr contained inthe frames which are
exchanged bythe data ink yer.
= When fame aves, the data tink layer process the frame
header and pases the contents ofthe fame payload field to
the network entity.
TPOUS wsnenenenne Transport ayer
Packets Network layer
Frames Data ink ayer
(6-394) Fig. 64.1
fechKnowledyé
Scanned by CamScanner Computer Networks (Comp-MU) 619 Transport Layer
4: Stream socket :
is designed for he conncton one tcl sch a
TCP. The TCP wes pir of seam ponte on x on
titer ends for comeing oe apaon program to
other across the Internet. a
2. Datagram socket
~ This type of socket is designed for the connectionless
protocol such as UDP.
UDP uses a pair of datagram sockets for sending a
‘message from one. application program to another
across the Internet.
3. Raw socket
= Raw sockets are designed for the protocols like ICMP or
OSPF, because these’ protocols do not use either steam
packets or datagram sockets,
Fig. 65.2 shows the three types of socket types.
[Se Aeticaton program
Ea Prsatedaeieee a
(6-02) Fig. 65.2: Type of sockets
6.5.2 Berkeley Sockets :
erin
F Wite short notes on : Berkeley sockot, :
(Dec. 09, Dec. 11, May 13, 6 Marks)
Q:2 Show the usage of the diferent _ socket
“programming primitives used for establishing @
connection between client and server.
ce (Dee, 13, 10 Marks)
2.3. White short note on Berkeley Socket, :
UNIX for TCP.
‘Create a new communication end point
Se] Primitive Meaning
No.
2_| IND Provide a local address a socket
LISTEN _ | Show willingness to accept connections
, [ACCEPT — | Block te caller as long as a connection
attempt does not arrive
5._| CONNECT _| Attempt to establish a connection
6,_| SEND Send data
7._| RECEIVE _| Receive data
| cLose | Release the connection
“The first four primitives ia the Table 6.5.1 are executed in the
same order by the server.
‘The SOCKET primitive creates a new end point and allocates
lable space fort within the transport entity
“The newly created sockets do not have addresses. These are
assigned using the BIND primitive.
‘The LISTEN primitive allocates space to queve the incoming
calls in case if several clients wish to connect at the same
time.
To block waiting for an incoming connection, the server
executes an ACCEPT primitive. When a TPDU requesting
for a connection arrives, the transport entity creates a new
socket and returns a file descriptor for it.
‘These were the primitives comesponding to server side. Now
let us consider the client side.
(On the client side also a socket needs to be created frst using
the SOCKET primitive, however the BIND is not required,
‘The CONNECT primitive blocks the caller and initiates the
connection process.
When it completes (Which is indicated by an appropriate
TPDU received from the server), the client process
unblocked and the connection is established.
‘After this both the sides can use SEND and RECEIVE
primitives to send and receive data,
In onder to release the connection, both sides have to execute
‘CLOSE primitive.
Steps followed for Socket Programming :
Server side
The steps followed for the socket programming ae as
follows :
Server creates a socket and checks for errors using SOCKET.
Assign adress tothe newly created socket using BIND,
Use the LISTEN to allocate space forthe queue whichis used
for the incoming calls,
Execute an ACCEPT for blocking the waiting incoming
connections.
TeehKeautedy
Scanned by CamScanner WB compurr Noworks (Comp: MU 14 Transport ay
Client side :
1, Create socket using SOCKET.
2. Use CONNECT to initiate connection process.
3" Establish the connection
6.5.3 Connectlonless Iterative Server
= Let us-now discuss connectionless, iterative client-server
‘communication using UDP and datagram sockets
= The server that uses UDP is usually connectionless iterative,
So the server serves one request a time.
= A-server gets the request received in a packet from UDP, it
processes the request and gives the response to UDP to send
ito the client.
= The server does not pay any attention to the other packets
= The other packets are stored in a queue waiting for the
service. They ar processed one by one,
~The server uses one single port for this purpose, the well
known port.
~All the packets ariving at this port will waitin line to be
served,
‘Server functions :
‘The server performs the following functions
(Create & socket : The server asks the operating system to
create a socket
2 Bind : The server asks the operating system to enter
information in the socket related to the server. This is called
as binding the server socket.
3. Repeat : The server repeats the following steps for infinite
numberof times.
(@) Receive request
(©) Process : The requests processed by the sever.
(© Send : The response is sent the client
CClents functions.
‘The client performs following factions
LL Create a socket : The client asks the operating system to
create a socket. There i no need of binding
2 Repeat: The clit repeats the following steps as long as it
has requests,
(2) Send + Cleat asks the operating system to send a
request
(b) Receive : Cleat asks the operating system to wait for
the response and deliver it when it has arived,
3, Destroy + When the client does not have any more requests,
asks the operating system to destroy the socket,
6.5.4 Connection Oriented Concurrent
Server:
- The oriented concurrent client serey
‘communication uses TCP and stream socket. The serves,
using TCP are normally of concurrent type. That means »
‘connection
server i serving many clients atthe same time
The type of communication is connection oriented. Once a
‘connection is established, it remains established until emir
stream of bytes is processed. After that the connection js
terminated.
= The server must have one buffer for each connection. The
bytes from the client are stored in buffers and hand
concurrently by the server.
~ In order to provide this type of server, the concept of pret
and child server suse.
Parent server :
= Apparent server isthe server running infinitely and accepting
connections from clients. It uses the well known por.
= After establishing a connection, the parent server creates 3
ew server called as a child server and an ephemeral post 9
allow the child server to handle the client.
‘Server function :
“The server performs following functions:
1. Create a socket : The server asks the operating system 10
crete a socket.
2 Bind + The server asks the operating system t0 cet
information in the socket.
3: Listen : The server asks the operating system to be past
and listen to the client which needs to be connected to this
server. Tis is because TCP is a connection oriented prot!
0 connection needs to be made before transfering the dst
4, Repeat : Tho server repeats the steps given below infil
(0) Create a child : When a child requests a connet"
the operating system creates a temporary oil
process and assign the duty of serving the client
the child. The parent process is then free for seine
to now clients.
Technaute
Scanned by CamScanner Computer Networks (Comp-MU)
oa *ognizing the dptons pate
sce sequence number is equied french es
recive so hep thos of toqeicc marke
ote host for a specie an
—— Pecifle amount of time but not
portant
problems :
= Buublishig a connection sound
: unds easy. But actually it is &
ver tcky jb. The problem occurs hen he net on
Jose store and duplicate packets, ”
= Theproblems canbe elaborated a8 follows
1, Due to congestion on a subnet, the acknowledgements
40 not get back in time from receiver to sender. So re-
‘transmission of each packet takes place
2. If the subnet uses datagrams inside and every packet
‘travels on a different route, then some of the packets
‘might get stuck ina trafic jam and take a long time to
‘The same connection getting re-established due to
duplication of packet.
= So the crux of the problem is existence of delayed duplicate
packets
+ The solution to this problem is to kill off the aged packets
that are still wandering on the network.
= We should ensure that no packet lives longer than some
predecided time.
~The packet lifetime can be restricted by using one of the
following techniques:
Restricted subnet design
2. Putting a hop counter in each packet,
3, Time stamping each packet.
6.7.2 Three Way Handshake Technique
AEE
pean i
‘Explain three way handshake technique in TOP:
ee (May 10, May 11, 10 Marks)
a
Explain the three protocol scenarios for establish
“a connection using a S-way handshake In TOP,
: (May 12, 10 Marke)
Show the. alifrent protocol scone ee
stablishing a connection using 3-way handshake In
ron (Dec. 13, 10 Marks)
jndshake technique In TCP.
if 16, 10 Marks)
Transport Lay
= The delayed duplicate packet problem can be solved by
using a technique called three way handshake.
= The principle of thee way handshake is shown in
Heat Hoa
(oF = Cennetion request
cK = Acknoiedgamer
Te Tie
(6-405) Fig. 6:7.2(a) : Three way handshake technique
Normal operation :
1, Host 1 chooses a sequence mumber x and sends a TPDU
containing the connection request (CR) TPDU to host 2.
2, Host 2 replies with a connection accepted TPDU to
acknowledge x and to announce its own sequence number y.
3, Host 1 acknowledges host 2 and sends the first data TPDU to
host 2.
Operation in the abnormal circumstances :
~~ Now let us see how the three way handshake works in
piesence of delayed duplicate control TPDUs.
= Refer Fig. 6.7.2(6). The first TPDU is a delayed duplicate
CONNECTION REQUEST from an-eld connection. The
HOST 1 does not know about it
Host t
Time:
Tine
(6-606 Fig. 6.7.2(b) : Response to an OLD duplicate
= Host 2 receives this TPDU and sends to host 1, a connection
accepted TPDU.
= But host 1 is not tying to establish any connection so it
sends a REJECT alongwith ACK = y.
Paes
Scanned by CamScanner uter Networks (Comp-MU)
‘So host 2 realizes that it was fooled by a delayed duplicate
‘and abandons the connection,
Duplicate CR and duplicate ACK:
‘Tris is another abnormal situation. Refer Fig. 6.7.2(¢) 0
understand this situation,
{(G-s07 Fig. 6.7.2) : Duplicate CR and duplicate ACK
‘This is the worst case in which delayed duplicates of both
‘connection request (CR) and acknowledgement (ACK) are
‘making rounds inthe subnet
Host 2 gets a delayed duplicate CR and it replies to it by
sending ACK. Note that host 2 has proposed a connection
witha sequence number y,
‘When the second delayed TPDU (duplicate) arrives at host 2
it understands that 2 has been acknowledged and not y. So it
understands that this too isan OLD duplicate:
6.7.3 Connection Release :
‘Any one of the two parties involved in data exchange can
close the connection,
But the problem is tha, when connection is terminated from
‘one end, he other party can continue to send data in the other
direction,
Hence, 10 ensure a proper connection release one has to
follow the steps given below.
Procedure to release a connection :
‘Refer Fig. 6.7.3 to understand this procedure,
Host | sends a connection release request to host 2,
Host 2 sends an acknowledgement to confirm the release
‘equest of host |
3. After this the connection is closed in one direction (no data,
{fom host 1 thos 2 but how 2can continue to send dao
6.8
Transport Lye,
‘When host 2 finishes sending his data, it sends a connec,
release request to host 1.
Host 1 acknowledges (confirms) the request made by joy »
and the connection is released from both ends. :
Host 2
=
Hoot
}
Connect rooase
roxio 102
|
Keknowledgemont
Connacton release
aquest 210 1
Hl
Tacleonledgement|
Time Tine
(6-409Fig. 6.73: Connection release
Releasing a connection is easier than establishing it The
are two styles of releasing a connection:
‘Types of connecting release :
1. Asymmetric release and
2, Symmetric release
A. Asymmetric release :
In asymmetric release, when one party stps
communicating the connection is broken. It is a2
abrupt release and it may lead to loss of data,
‘Symmetric release :
‘Symmetric release treats the connection as two sears
‘unidirectional connections and ia order to release
connection each side must release the connection 00 iS
ide.
‘Symmetric release is as shown in Fig. 6.73 and ther
‘no loss of data with the symmetrical release
The Internet Transport Protocols
(TCP and UDP) :
‘The Internet as two main protocols i the transpor HY
One of them is connection oriented and the otes °
supports the connectionless service,
‘TCP (Transmission Control Protocol) is a conectot
oriented protocol and UDP (User's Data Protocl) is
connectionless protocol
‘UDPis basically just IP with an additional short Kade
wreett
Scanned by CamScanner WF computer Networks (Com
69
Uy
User Datagratn Protocol (UDP) :
= The User Datagram Protocol is a very simple protocol, It
adds litle to the basic functionality of IP. Like TP, it i
unreliable, connectionless protocol.
= You do not need to establish a connection with a host before
exchanging data with it using UDP, and there i
‘mechanism for ensuring that data sent is received,
= A onit of data sent using UDP is called a Datagram. UDP
adds four 16-bit header fields (8 bytes) to whatever data is
sent,
= These fields are : a length field, a checksum field, and source
and destination port numbers. “Port number”, inthis context,
represents a software port, nota hardware port.
= The concept of port numbers is common to both UDP and
‘TCP. The port numbers identify which protocol module sent
(ors to receive) the data.
Most protocols have standard ports that are generally used
for this. For example, the Telnet protocol generally uses port,
23. The Simple Mail Transfer, Protocol (SMTP) uses port 25.
The use of standard port numbers makes it possible for
clients to communicate with 2 server without fist having to
establish which port to use.
= The port number and the protocol field in the IP header
duplicate each other to some extent, though the protocol field
is not available to the higher-level protocols. IP uses the
protocol field to determine whether data should be passed to
the UDP or TCP module.
= UDP or TCP use the port number to determine which
application-layer protocol should rceive the data
= Although UDP isn’t reliable, itis stil a preferred choice for
‘many applications, It is used in real-time applications like
[Net audio and video where, if data is lost, its better to do
‘without it than send it again out of sequence. Iti also used
by protocols like the Simple Network Management Protocol
(SNMP).
Relationship with other protocols :
= The relationship of UDP with the other protocols and layers
‘of TCP/IP suite is as shown in Fig. 6.9.1. As shown, UDP is
focated between IP and application layer. It therefore works
8 an intermediary between application program and the
network layer
SMTP, FTP, ONS, HOP
[ac sctr, Tor, bP
TP, ARP, IGMP, [OMP
‘Application layer
‘Transport layer
"Notwork layer
; eee bem nce
is eSsog [Pres e
etme UDP and te roa
(Gang) Fig, 69.1 : Relation
6.25,
Transport Layer,
6.9.1
= Being a transport layer protocol, the UDP has the following
responsibilities
1. To create a process to process communication, UDP
ses port numbers to accomplish this.
2. To provide control mechanisms at the transport layer,
UDP does not provide flow control or
acknowledgements, It provides error detection, The
erroneous packet is discarded.
3, UDP does not add anything to the services of IP
except for providing process to process
6.9.2 Advantages of UDP
= UDP, despite alts simplicy and poweressnes is still used
because i offers the fllowing advantages
1. UDP has minimum overheads
2, UDP can be eaily‘used if the sending process isnot
too bothered about reliability
3. UDP reduces interaction between sender and receiver.
Responsibilities of UDP :
6.9.3. User Datagram :
= User Datagram Protocol (UDP) provides a connectionless
packet service that offers unreliable “best effor” delivery.
‘This means thatthe arrival of packets isnot guaranteed, nor
is the corect sequencing of delivered packets.
= Applications that do not require an acknowledgement of
receipt of data for example audio or video broadcasting uses
pr.
UDPiis also used by applications that typically transmit small
amounts of data at one time, for example, the Simple
[Network Management Protocol (SNMP).
= UDP provides a mechanism that application programs use to.
send data to other application programs. UDP provides
protocol port numbers used to distinguish between multiple
‘programs executing ona single device.
That is, in addition to the data sent, each UDP message
contains both a destination port number and a source port
number. This makes it possible for the UDP software at the
destination to deliver the message to the correct application
program, and forthe application program to send a reply.
= UDP packets aro called as user datagrams. They have a
fixed-size header of &-bytes. The format of user datagram is
as shown in Fig. 69.2
— Bbyt0s 4
[ies] oe]
Destination port
umber 16 bits
Total ongth
seks
(G-026Flg, 69.2 : User datagram format
‘Source port
number 16 bits
Technet
Hester
Scanned by CamScanner Computer Networks (Comp-MU) 6.26 Transport Layer
~The UDP header is divided imto the following four 16-bit
fields :
1. Source port 3. Total length
2 Destination port 4, Checksum.
Source Port Number :
= Source ports an optional el, when meaning, itindictes
the port of the sending process, an may be assumed to be
‘the port to which a reply should be addressed ia the absence
‘of any other information. IF not used, a value of zero is
inserted.
= Thisis a 16 bit eld. That means the port numbers can range
from 0t0 65,535,
= If the souee host is a client, means if a client is sending @
request using UDP, then generally a ephemeral (temporary)
Port number is requested by the process and choten by the
‘UDP.
— Ifthe source host isa server thai means ia server is sending
response message, most the well known port number is
~ sed.
Destination Port Number :
~The destination port number also isa 16 bit number and this
port number is used bythe press running on the destination
host.
= If the destination host isa server that means if client is
sending a request to it, then a well known port number is
sed in most cases.
= However if the destination host is client than means if a
server is sending its response to it, then the chosen port
‘number is generally an ephemeral port number.
Length
= Its also a 16 bitfield which is used for defining the total
length ofthe UDP datagram including header as wel as data.
Due to 16 bit length it can define a total tength of the
atagram upto 65,535 byes
= However practically the total length of a UDP datagram is
rmuch smaller than 65,535 bytes. This is because the UDP
datagram isto be stored in an TP datagram which tet has 9
Iength of 65,535 bytes.
= The length field’ in the UDP. datagram is scaly not
necessary, because this UDP datagram is "actually
‘encapsulated in an IP datagram and the IP datagram has is
own length field
= So without using the length field in UDP datagram, we can
‘obtain the length ofthe UDP datagram as follows
UDPiength = IP length IP header length
= Note that while delivering the UDP datagram to UDP layer
UDP Checksum :
he integrity (i.e. to detect errors) ofthe
‘This is sed to verify th
‘on a “pseudo header”
UDP header, The checksum is performed
consisting of information obtained from the IP beader (Source and
destination address) as well a the UDP header.
6.9.4 UDP Pseudo Header +
= The purpose of using a pseudo-header is to verify that
UDP packet has reached its correct destination.
fic machine and 3
= The correct destination consists of a. speci
specific protocol port number within that machine.
°
ra
[es
(6-0 Fig. 69.3 : UDP pseudo header
= The’ UDP header itself specifies only the protocol pos
‘number. Thus, to verify the destination, UDP on the sending
‘machine computes a checksum that covers the destination IP
‘address as well asthe UDP packet.
= At the ultimate destination, UDP software verifies the
checksum using the destination IP address obtained from the
header of the IP packet that carried the UDP message
~ If the checksum agrees, then it must be true thatthe packet
has reached the intended destination host as well as the
correct protocol port within that host.
User Interface :
= A user interface should allow the creation of new receive
ports, receive operations on the receive ports that retum the
data octets and an indication of source port and source
address, and an operation that allows a datagram to be sent
specifying the dats, source and destination ports and
‘addresses tobe sent.
IP Interface :
= The UDP module must be able to determine the source and
destination Internet addresses and the protocol field from the
Internet header,
= One possible UDPP interface would retum the whole
Intemet datagram including the entire Intemet header is
tespons to eciv operation, Such an interac woul io
allow the UDP to passa full Internet ‘complete with
header tothe IP send, aoe
the IP software dops the IP header. = The IP would verify cenain elds for consistency and
‘compute the Internet header checksum
= Teehknvatest
Scanned by CamScanner protocol Application
and the Trivial File Transfer.
protocol Number :
= This #8 protocol 17 (21 octal) when used
eel the Iniemet
‘The dump of a UDP header in hex
Terra ne UDP Header in hoxadacial
BC82000D002B001D
bia the following from it
1. Source port number
2, Destination port number
3. Total length
4. Length of the data,
5. Packet direction.
6. Name of client process.
soln. :
The standard format of UDP header has been shown in
Fig. P.69.1.
f= B bytes —o}
Which shows that for well known port number 13, the
corresponding client process is “Daytime”,
6.10 UDP Services :
= In this section we ate going to discuss the following
{mportant services provided by the UDP.
Process to process communication.
Connectionless services,
Flow contr.
Error control.
Checksum.
Congestion conto.
Encapsulation and decapsul
Queuing
Multiplexing and demultiplexing.
6.10.1 Process to Process Communication
= We have already discussed the process to process
‘communication ina general sense, earir inthis chapter.
2
a
4
5.
6
1
8
8.
= UDP also does it with the help of sockets which is a
combination of IP address and port numbers. Table 6.10.1
shows diferent port aumbers used by UDP.
‘Some of these ports can be used by UDP as well as TCP.
Table 6.10.1: Well known ports used with UDP.
Protocol | Description
7 | Bebo | The received datagram is echoed back to
-—— 22 bis +t cele
9 | Discard | Any received datagrams discarded.
(cana Fig. P- 69.1 : UDP header format Gill tues increas
= Therefore we can splithe given UDP headerin equal pars || 13 [Daytime | Rerum he day andthe cunt tine.
as follows 17 | Quote | Return the quote of he day.
BC82 000D 0028 0010, 19 | Chargen | To retuma string of characters.
Scirce pot ] L Lecnecteum |{s3 | Nanesener | Donia Nene Serie NS).
ee Length 67 | BOOTRS | This is te server port to download the
‘number bootstrap information,
ae 68 | BooTeC | This is the client port to download
1. Source port umber = (BC82)4 wns. bootstrap information
2. Destination port number = (000D), veas. 169 [7 Tete | Trivial File Transport Protocol,
3. Tal engthof UDP packet =(0028)«=3)ete8 UT inn [rc | Remo Proved Cal
123 [_NTP_| Nework Time Protocol,
4. Length of data = Toa length Length of hehesters | 161 | Sage _| simple Nework Manageteat Protocol
43-8 = 35 byles wAm 1 Trea | swMP | simple Network Management Prooco
5. Destination port aumber is (000D)6= (13) np) 3
~ this a wellknown port Hence the diretion of UDP packet
save is from cient to server.
6 The client process can be obtained from Table 6.10.1
ates
Scanned by CamScanner “Computer Networks (Comp-MU)
6.10.2 Connectionless Services :
~ As UDP is a connectionless, unreliable protocol, each user
datagram sent using UDP isan independent datagram,
Different user datagrams sent by the UDP have absolutely no
relationship between them, This is true even for those
datagrams which are originating from the same process and
being sent to the same destination, The user datagrams do
not have any number,
~ Also the connection establishment and release are not at all
‘equired, So each datagram is free to travel any path.
= Only those processes which are sending very short messages
‘can successfully use the UDP.
6.10.3 Flow and Error Control :
~ Being a connectionless protocol, UDP is a simple, unreliable
protocol. It does not provide any flow control, hence the
receiver can overflow with incoming messages,
= UDP does not support any other error control mechanism,
‘except for the checksum.
There: are no acknowledgements sent from destination to
sender. Hence the sender does not know if the message has
reached, lost or duplicated. If the receiver detects any error
using the checksum, then that particular datagram: is
discarded.
6.10.4 Checksum:
= The calculation of checksum for UDP is different than that
for IP. In UDP the checksum is calculated by considering the
following three sections
1. A pseudoheader
2. The UDP header.
3. The data coming from the application layer.
~The checksum in UDP is optional. That means the sender
‘ean make a decision of not calculating the checksum. If s0,
then the checksum fields filled with all zeros before sending
the UDP packet
= In case if the calculated checksum is all zeros (when the
sender decides to send checksum) then an all 1 checksum is
sent.
= This solution works without any problem because, a
checksum will neverhave anal I value
6.10.5 Congestion Control
= UDP does not provide any congestion contol. It assumes
that the UDP packets being small, will not create any
congestion.
— But this assumption may not always be correct
6.10.6 Encapsulation and Decapsulation
~The UDP encapsulates and decapsulates messages in an IP
datagram in order to exchange the message between two
‘communicating processes.
628
‘Transport Layer
= This is as shown in Pig, 610.1. We will discuss the wo
processes separately.
‘Sencirg process
oP satan
(b) Decapsulation
(Ga02y Fig. 610.1
(a Encapsulation
Encapsulation :
Refer Fig. 6.10.1(2). The message produced by a process is
to be sent with the help of UDP. The process passes the
-mestage and two socket addresses alongwith the length of
data to UDP.
= UDP receives this data and adds the UDP header to it as
shown. This is called as UDP datagram which is passed to IP
with the socket address
= TPadds its own header to UDP datagram as shown. It enters
value 17 into the protocol field. This is an indication chat
UDP is being used. The IP datagram is then passed on tothe
ata lnk layer.
= The DLL adds its own header and possibly a trailer to create
frame and sends itt the physical layer.
= Finally te physical layer converts these bits into electrical or
optical signals and sends them tothe destination machine.
Decapsulation :
= Refer Fig. 6.10.1() for understanding of the decapsulation
process. The encoded message arrives at the destination
physical layer where it decodes the electricaVoptical signals
into bits and passes them to the DLL
~The DLL checks the data using header and trailer. The
header and trailer are discarded if no errors are found, and the
datagram is passed to IP.
~The IP cartes out its checking to find the errors and if none
are found, the datagram is passed on to UDP, after dropping
the IP header. %
= The datagram from IP to UDP also contains the sender and
receiver IP addresses, This entire user datagram is checked
by the UDP with the help of checksum,
= there is no error detected, then the UDP header is dropped
and the application data plus senders socket address are
hhanded over tothe process.
Scanned by CamScanner Computer Networks (Comp: MU!
= The process can use this senders socket address if it wants to
respond to the message received.
6.10.7 Queuing
= The queues in UDP are related with ports a6 shown in
Fig. 6102.
ay te cont Day time sone
gory Incoming
queue ‘a
(6-426 ig. 6.10.2 : Queues in UDP
= A process starts atthe client site by requesting a port number
from the operating system. In some implementations both
incoming and outgoing queues are created ia association with
cach process.
. Every process gets only one port number and hence it can
‘create one outgoing and another incoming queve. The queues
function only when the process is running. They are
destroyed as soon asthe process is terminate.
=. The client process uses the source port number mentioned in
the request to send message to its outgoing queue.
= = UDP removes the queue messages one by one by adding the
‘UDP header and delivers them to IP.
~ Ifthe outgoing queue overflows, then operating system tells
that client process to wait before sending the next message.
"When the client receives a message, UDP checks if the
incoming queue has been created or not If the queue has
‘een created, then the UDP sends the received datagram to
the ead of the queve.
= Tf the queue is not present then UDP will simply discard the
user datagram. If the incoming queue overflows, then UDP
discards the user datagram and arranges’ to send the port
unavailable message to the server.
=, The mechanism to create the server queue is different. The
server creates the incoming and outgoing queues using its
‘well kaown port as soon as it starts running, The queues exist
as long asthe server is running,
~ * When a message is received at the server, the UDP checks if
the incoming queue has been created or not.
Ifthe queve is not present, the UDP discards the user
datagram. If the queue is present then UDP sends the
datagram atthe end of the queue.
~~" If the incoming queue overflows, then UDP drops the user
datagram and arranges to send the port unavailable message
tothe client
Transport
= When the server wants to send a message to client it sends
that message tothe outgoing queue. These messages are then
removed one by one after adding the UDP header. They are
delivered 1 IP.
= Ifthe outgoing queue overflows then the operating system
will ask the server to wait before it sends the next message.
6.10.8 Multiplexing and Demuttiplexing :
= We have discussed the general principle of multiplexing and
“demultiplexing in the taneport layer.
= Now let us see how to apply the same principle to UDP.
Imagine that a host is running a TCP/IP protocol suite and
that there is only one UDP and a number of processes which
Would lke to use the services of UDP.
UDP handles such a situation by using the principle of
‘multiplexing and demultiplexing as shown in Fig. 6.103.
Procosses
Processes
] El
PN
(U0)
(G-202» Fig. 6.10.3 : Multiplexing and demultiplexing
Multiplexing :
= At the sending end, there are several processes that are
interested in sending packets. But there is only one transport
layer protocol (UDP or TCP), Thus it is a many processes-
‘one transport layer protocol situation.
Such a many-to-one relationship requires multiplexing.
=" The UDP first accepts messages from different processes.
‘These messages are separated from each other by their port
‘numbers. Bach process has unique port number assigned to
it
= Then the UDP adds header and passes the packet to IP as
shown in Fig, 6.103.
Demuttiplexing :
- At the receiving end, the relationship is one as to many, So
‘we need a demultpiexer,
'~ First the UDP layer receives datagrams from the IP.
= The UDP then checks for error’ and drops the header to
‘obtain the messages and delivers them to appropriate process
‘based on the port number.
ee
Scanned by CamScanner Computer Networks (Comp-MU)
6.10.9 Comparison of UDP and Generic
Simple Protocol :
= In this section we will compare UDP with a
cconnectionles transport layer protocol
= The only difference between the two is that the UDP
‘provides an optional checksum,
= If the checksum is added to the UDP packet then at the
destination, the receiving UDP can check the packet for any
‘ctor withthe belp ofthe checksum.
~ If any eror is detected, the receiving UDP will discard that
packet, without sending any feedback tothe sender.
imple
6.11 . UDP Applications :
Despite being connectinless, unreliable, no flow control, no
error control, UDP is sill prefered for some applications.
This is because UDP has some advantages too. An
application designer has to sometimes compromise between
advantages and drawbacks t get the optimum.
— "Here we will discuss some important features of UDP that
_are useful in designing an application program.
6.12 UDP Features :
6.12.1 Connectionless Service :
+The feature of UDP is that its aconnectionless protocol and
tat each UDP packet is independent from the other packets,
ean be considered as an advantage or a. disadvantage
depending on the requirements of an application.
~ Iman application, if we want to send only short messages to
server and receive short messages from the serve. Then the
above mentioned feature becomes an advantage,
= The feature of being connéctionless is an advantage if request
‘and respond each can fit in one single user datagram,
= The overhead (number packets to be exchanged) required to
establish and close a connection is zero in case of UDP. This
can be a very important advantage for some applications.
Similarly the delay involved with the connectionless delivery
is very short. as compared to that with the connection oriented
The relationship between TCP and IP is very interesting
Each TCP message gets encapsulated or inserted in an IP
‘atagram and then this datagram is sent over the Intemet to
the destination.
~ BP wansports this datagram from sender to destination,
without bothering about the contents ofthe TCP message.
= Atthe final destination the IP hands over the message tothe
{TCP software running on th destination computer.
= acts tke postal service and transfers the datagrams from
‘one computer tothe other.
Thus TCP deals withthe actual data to be transfered and TP
takes care of transfer of that data.
= Many applications such as FTP, Remote login TELNET etc
Leveeeing dat TCP waar onthe ending compte
= The TCP software acts as a muliplexer at the sending
‘computer, Tt receives data from various applications,
fhuldplenes the data and hands it over tothe IP software at
the sending end as shown in Fig 6.13.1.
~ BP adds its own header to this TCP packet and creates an TP
‘packet out oft Then this packet is sent to its destination
: ss wil take place
~ Atte destination exactly opposite proce
‘The IP software hands over the multiplexed data to the TCP
software.
~The Top
demultipenes the multiplex
comesponding applications a.
software at the destination computer then
od data and gives it 10 the
shown in Fig. 6.13.1.
‘Arpleaton pagans ‘Aeration programs
‘heonder the gecnaton
TF Tor
uvtotoan (cumiisonen)
1° he
Payal Pryseal
eter ra
Qo
(¢-14m Fig, 6.13.1 : Maliplexing and demultiplexing using
Tc
6.13.2 Ports and Sockets :
1. > Ports:
= Applications running on different hosts communicate with
‘TCP with the help of ports. Every application has been
allotted a unique 16 bit number whichis known as a port.
= When an application on one computer wants to communicate
‘using a TCP connection to another application on some other
‘computers these ports prove tobe very helpful
= Letan application A on host 1 wants to communicate with an
application B on host 2. So the process takes place as shown
in Fig. 6132 and explained below.
‘options
(6-167 Fig. 6.13.2 : Use of port numbers
= Application A running on computer 1 provides the IP address
‘of computer 2 and the port number comesponding to
application B as shown in Fig, 6.13.2.
Scanned by CamScanner ‘Computer Networks (Comp-MU)
= Computer 1 communicates with computer 2 using the IP
address and computer 2 uses the port number to direct the
‘message to application B,
2 Sockets:
A port is a 16 bit unique number used for identification of a
single application,
Bot socket address or simply socket would identify the
combination of the IP address and the port number
602
6.14.1 Process to Process Communication
‘The TCP uses port numbers a transport layer addresses,
‘Table 6.14.1 shows some well known port numbers used by
Te,
= Note that if an application ean use both UDP and TCP, the
same port number is assigned t this application.
‘Table 6.14.1 : Well known ports used by TCP
For example if the IP address = 193,25.10.12 and the port Toprol | Description
Ser eee a a et SS
following socket address. i pes
Sain
a >| er areca
‘Port 13 Daytime | Sends the date and the time
Seine |] ES sad
oe tetas
(100 Fig, 6133 20 | FTP, Data | Fite Transfer protocol for data
~ Soaprir of sockets is required to identify a TCP connection] 24 FIP, Control _| File Transfer protocol for control
‘between two applications on two different hosts. These (wo ry - a ok
socket addresses specify the end points of the connection as bs re
nr Se eee sen
= St Pewee
See
= ole er
een ee
(G-1s6 Fig. 6.134 : Source and destination port numbers
= Generally the server port numbers are known 2s the well
known ports. Some of the well known port numbers have
already been mentioned for UDP and TCP earlier in this
chapter.
= Multiple TCP connections between diferent applications or
same applications on two hosts exis in practice. Here the IP
addrestes ofthe two hosts are same but the port numbers are
éifferent.
= The communieation using port numbers is illustrated in
Fig. 6.134,
6.14 TCP Services:
Following are some of the services offered by TCP to the
processes atthe application layer
Stream delivery service
Sending and receiving buffers
Bytes and segments
Full duplex service
‘Connection oriented vervice
Reliable service,
Process to process communication
SAR a eDe
6.14.2 Stream Delivery Service :
= TCP is a stream oriented protocol. The sending process
delivers data in, the form of a stream of bytes and the
receiving process receives it in the same manner.
TCP creates @ working environment in such way thatthe
sending and receiving processes seem to be connected by an
imaginary “tube” as shown in Fig. 6.14.1.
This is called as stream delivery service.
‘Sending procoss Receiving process
Imaginary tube cay
thodata groom?
(-sa0Flg. 6.1441: Stream delivery service
6.14.3 Sending and Receiving Buffers
~The sending and receiving processes may not produce and
receive data atthe same speed,
TechKnowledgé
Scanned by CamScanner Computer Networks (Comp-MU
ese TP redseton or
ong dan tbo en
‘There are two types of buffers used in each diecton: “s
i eeeeincee ter
A buffer can be implemented
Receiving buffer.
by using a circular array of 1
byte locations as shown in Fig 6.14.2.
sending
process
Rocahving
Noxt byte _Noxttyte
Tobesent tobe
recelved
6-33 Transport
‘The Segments are then inserted in an IP datagram and
transmitted. The entire operation is transparent to the
receiving process.
‘The segments may be received out of order, lost or corrupted
when it reaches the receiving end.
Fig. 6.143 shows the creation of segments from the bytes in
the buffers.
‘Sending TOP
Recoling TOP
D1 Empty locations
D1 Bytes cont but acknowledgement
Ienat received
BB Bytestobe sont
(G-aFig. 6.142 : Sending and receiving bsfers
Fig. 6.142 shows the direction of movement of data. The
sending buffer has three types of locations
1, Empty locations
Emp locations
2, Locations containing the bytes which have been sent
bat not acknowledged, These bytes are kept in the
baffer till an acknowledgement is received
3, The locations containing the bytes to be sent by the
sending TCP.
In practic, the TCP may be able to send only a part of data
Which is to be seni, doe to slowness of the receiving process or
congestion in the network.
‘The buffer atthe receiver is divided into two pars :
1. ‘The part containing empty locations.
2. The part containing the received bytes which can be
‘consumed by the sending process.
6.14.4 Bytes and Segments
Buffering i used to handle the difference between the speed
of data transmission and data consumption.
But only buffering is not enough. We need one more step
before sending the data.
‘The IP layer, which provides service to TCP, has to send data
{n the form of packets instead of stream of bytes
[At the transport layer, TCP group a numberof byes to form
‘packet called a segment.
‘A header is added to each segment for the purpose of
exercising control.
‘Sending TOP.
(G2 Fig. 6143
‘The segments are not of the same size. Each segment can
camry hundreds of bytes.
6.14.5 Full Duplex Servic
= TCP offers full duplex service where the data can flow in
both the directions simultaneously.
Each TCP will then have a sending buffer and receiving
buffer. The TCP segments can travel in both the directions,
‘therefore TCP provides a full duplex service.
6.14.6 Connection Oriented Service :
= TCP is a connection oriented protocol. When process — 1
wants to communicate (send and receive) with another
process (process - 2), the sequence of operations is as
follows
L. TCP of process ~ 1 informs TCP of process ~ 2 and.
create a connection between them.
2, TCP of process ~ 1 and TCP of process ~ 2 exchange
