CONGESTION of packets from the user process in an even flow

occur in a computer network when the resource of packers onto the network. Conceptually. each
demands exceed the capacity Packets may be host is connected to the network by an interface
lost due to too much queuing in the network. Containing a leaky bucket, that is, a finite
During Congestion the network throughput may internal queue. If a packet arrives at the queue
drop and the path delay may become very high when it is full , the packet is discarded
Congestion in a network may occur if users send
data into the network at a rate greater than
allowed by network resources. for example,
Congestion may occur because the switched in a
network have a limited buffer size to store
arrived packets before processing
Causes of Congestion
1: Unpredictable statistical fluctuation of traffic In other word, if one or more processes within
flows the host try to send a packet when the
2: faults conditions within the network maximum number is already queued, the new
3: Slow processor speed. if the router's CPU packet is unceremoniously discarded. This
speed is very low and performing tasks like arrangement can be built into the hardware
queuing buffer, tables updating etc. queries are interface or simulated by the host operating
built up, even though the line capacity is not system. It was first proposed by Turner (1986)
fully utilized and is called the leaky bucket algorithm. In fact
4: Inefficient control policies it is nothing other than a single-server queuing
5: Bandwidth of the links is important in system with constant service time. This
Congestion. The links to be used must be of high mechanism turns an uneven flow of packets
Bandwidth to avoid Congestion from the user processes inside the host into an
even flow of packets onto the network.
Smoothing out bursts and greatly reducing the
chances of congestion

TYPES OF CONGESTION CONTROL

ALGORITHMS
Congestion in a frame relay network is a Implementing the original leaky bucket
problem that must be avoided because it algorithm is easy. The leaky bucket consists of a
decreases throughput and increases delay finite queue. When a packet arrives, if there is
1: Leaky Bucket Algorithm: If there is a hole at room on the 4ueue it is appended to the queue
the bottom of a Bucket, then no matter at what otherwise, it is discarded. At every clock tick,
rate the Bucket is filled up, the water leaks out one packet is transmitted (unless the queue is
drop by drop at a constant rate from the hole. empty).
Each host is connected by an interface that has This arrangement can be simulated in the
finite queue acting like leaky Bucket. operating system or can be built into the
When a packet comes to a host with the queue hardware. Implementation of this algorithm is
full, it is discarded. The host 1s allowed to put easy and consists of a finite queue. Whenever a
one packet per clock tick info the network. This packet arrives, if there is room in the queue it is
can he enforced by the interface card or by the queued up and if there is no room then the
operating system. This converts an uneven flow packet is discarded
2: Token Bucket Algorithm *It allows large bursts to he sent at faster rate
This algorithm allows bursts for short after that constant rate.
transmission while making sure that no data is *It saves token to send large bursts.
lost. In contrast to the Leaky Bucket algorithm, Leaky Bucket Algorithm
not the data that is to be send but tokens are *Token independent
queued in a time-depended queue. One token is *If bucket is full packet or data is discarded.
needed to send a single portion of data *Packets are transmitted continuously.
Implementations contain a token counter that is *It sends the packet at constant rate.
incremented on every time interval, so that the *It does not save token.
counter grows over time up until a maximum QUALITY OF SERVICE
counter value is reached. The token counter is is defined as something flow seeks to attain
decremented by one for every data portion A Stream of packets from a source to
sent. When the token counter is zero no data destination is called flow. In a connection
can be transmitted. oriented network all packets belonging to a flow
For many applications it is better to allow the follow the same route, in a connection-less
output to speed up somewhat when a larger service they may follow different routes
burst arrives than to lose the data. Token The needs of each flow can be characterised by
Bucket algorithm provides such a solution. primary parameters vi, reliability, delay, jitter
and bandwidth. together these determine the
QoS(Quality of Service) the flow requires
QoS defines a set of attributes related to the
performance of the connection for each
connection the user can request a particular
attributes.
flow Characteristics
1: Reliability: Reliability is a characteristic that
flow needs. lack of reliability means losing a
Main steps of this algorithm can be described as packet or acknowledgment which entails
follows: retransmission. However, the sensitivity of
* In regular intervals tokens are thrown into the application programs to reliability is not the
bucket. same
* The bucket has a maximum capacity. 2: Delay: Source-to-destination delay is another
* If there is a ready packet, a token is removed characteristic Again application can tolerate
from the bucket, and the packet is send. delay in different degree. In this case,
* If there is no token in the bucket, the packet telephony, audio conferencing, video
cannot be send. conferencing and remote login need minimum
delay, while delay in file transfer or e-mail is less
important
3: Jitter: Jitter is the variation in delay for
packets belonging to the same flow. For
example, if four packets depart at times 0,1,2
and 3 and arrive at 20,21,22 and 23, all have the
same delay, 20 units of time. Jitter is defined as
the variation in the packet delay. High jitter
means the difference between delays is large;
low jitter means the variation is small
Token Bucket Algorithm 4: Bandwidth: Different applications need
*Token dependent different bandwidths. In video conferencing one
*If bucket is full token are discarded, but not need to send millions of bits per second to
the packet. refresh a colour screen while the total number
*Packets can only transmitted when there are of bits in an e-mail may not reach even a million.
enough token.
QoS Attributes every authorised member of which could
1: User Related Attributes: These are related to communication with every other authorised
the end user in the sense that they define how member directly or indirectly
fast a user wants to send/receive data Shows Interconnection Of 2 WAN and 1 LAN
The Attributes are negotiated and defined at
the time of the contact between the user and
the network service provider
*Sustained Cell Rate(SCR): This is the average
cell rate over a period of time, which could be
more or less the actual transmission rates, as
long as average is maintained
*Peak Cell Rate(PCR): This is the maximum
transmission rate at a point of time
*Minimum Cell Rate(MCR): This is the minimum
cell rate that the network guarantee a user
*Cell Variation Delay Tolerance(CVDT): This is a
unit of measuring the changes in cell
Types or Internetworking
transmission times 1: Concatenated Virtual Circuit: In the
2: Network Related Attributes: These Attributes concatenated virtual circuit model, shown in
defines the chara of a network figure connection to a host in a distant network
*Cell Loss Ratio(CLR): This Attribute define the is set up in a way similar to the way connections
fraction of cells lost/delivered too late during are normally established. The subnet sees that
transmission the destination is remote and builds a virtual
*Cell Transfer Delay(CTD): This is the average circuit to the router nearest the destination
time required for a cell travel from the source to network. Then it constructs a virtual circuit from
destination that router to an external "gateway"
*Cell Delay Variation(CDV): This is the difference (multiprotocol router). This gateway records the
between maximum and minimum values of CTD existence of the virtual circuit in its tables and
*Cell Error Ratio(CER): This parameter defines proceeds to build another virtual circuit to a
the ratio of cells that contain errors router in the next subnet. This process
continues until the destination host has been
INTERNETWORKING
reached Once data packets begin flowing along
As the computer got smaller, cheaper and yet
the path, each gateway relays incoming packets.
more powerful. more and more organisation,
Converting between packet formats and virtual
companies and people began having their own
circuit numbers as needed. Clearly, all data
private network, even Internetworks in case of
packets must traverse the same sequence of
large organisations
gateways, and thus arrive in order
Most of them wanted to join the rest of
information world by further connecting to the
internet. some of organisations used internet as
a vehicle of communication between their
remotely located private network/
Internetworks
All of these development saw the
Internetworking technology to evolve as an
important technology. an Internetworking is a
collection of individual networks, connected by The essential feature of this approach is that a
intermediate networking devices, that function sequence of virtual circuits is set up from the
as a single large network. there are various source through one or more gateways to the
types of networks like LAN,WAN,MAN destination. Each gateway maintains tables
An Internetwork may also be defined as a telling which virtual circuits pass through it.
network of computer communication network
2: Connectionless internetworking: The INTERNET PROTOCOL (IP)
alternative internetwork model is a the IP is a datagram-oriented protocol, treating
datagram model, in this figure datagram from each packet independently. Also Internet
host 1 to host 2 are shown taking different Protocol makes no attempt to determine if
routes through the internetwork packets reach their destination or to take
In this model, the only service t11e network corrective action if they do not. Internet
layer offers to the transport layer is the ability Protocol provides the following functions:
to inject datagram into the subnet and hope for 1: Addressing, 2: Fragmentation
it will get to the destination. There is no notion 3: Packet timeouts
of virtual circuit at all in the network layer. let Internet Protocol (IP) is a network layer (Layer
alone a concatenation of them. This model does 3) protocol chat contains addressing
not require all packets belonging to one information and some control information that
connection to traverse the same sequence or enables packets to be routed. Along with the
gateways. transmission Control Protocol(TCP), IP
A routing decision is made separately for each represents the heart of the Internet protocols.
packet, possibly depending on the traffic at the IP has two primary responsibilities
moment the packet is sent. This strategy can 1: Providing Connectionless, best effort delivery
use multiple routes and thus achieve a higher of datagram through an internetwork
bandwidth than the concatenated virtual circuit 2: Providing fragmentation and reassembly of
model. On the other hand, there is no datagram to support data links with different
guarantee that the packets arrive at the maximum transmission unit(MTU) sizes
destination in order, assuming that they arrive IP Packet Format
at all

1: Version: Indicates the version of IP currently

used.
2: IP Header Length (1HL): Indicates the
Problems in Connectionless Internetworking datagram header length in 32-bit words.
1: Conversion: If each network has its own 3: Type-of-Service: Specifies how an upper-layer
network layer protocol, it is not possible for a protocol would like a current datagram to be
packet from one network to transit another handled, and assigns datagram various levels of
one. Multiprotocol routers tries to translate importance.
from one format to another. But such 4: Total Length: Specifies the length, in bytes, of
conversions will always be incomplete and often the entire IP packet, including the data and
move to failure unless the two formats are close header.
relatives with the same information fields For 5: Identification: Contains an integer that
this reason conversion is rarely attempted. identifies the current datagram. This field is
2: Addressing: Imagine a simple case: a host on used to help piece together datagram
the Internet is trying to send an IP packet to a fragments.
host on an adjoining OSI network. The OSI 6: Flags: Consists of a 3-bit field of which the
datagram protocol. Problem is that IP packets two tow-order (least-significant) bits control
all carry the 32-bit Internet address of the fragmentation.
destination host in a header field. OSI hosts do 7: Fragment Offset: Indicates the position of the
not have 32-bit Internet addresses. They use fragment's data relative to the beginning of the
decimal addresses similar to telephone numbers data in the original datagram
8: Time-to-Live: Maintains a counter that
gradually decrements down to zero, at which
point the datagram is discarded.
9: Protocol: Indicates which upper-layer
protocol receives incoming packets after IP
processing is complete. IP Address Classes
10: Header Checksum: Helps ensure IP header A class is used to recognise the part of network
integrity. address and node address given in an IP
11: Source Address: Specifies 1he sending node. address. There are 5 classes associated with IP
12: Destination Address: Specifies the receiving addresses. A,B,C,D,E where A,B,C are used for
node. commercial purpose
13: Options: Allows IP to support various
options, such as security.
14: Data: Contains upper-layer information.
IP ADDRESSING
As with any other network-layer protocol, the IP
addressing scheme is integral to the process of
routing IP datagram through internetwork. Each
IP address has specific components and follows
a basic format These IP addresses can be
subdivided and used to create addresses for sub
networks.
1: Network Number: It identifies a network and 2: Classless Addressing: There were certain
must be assigned by the internet network problems with classful addressing such as
information centre(InterNIC) if the network is address depletion and less organization access
to be part of internet. an ISP can obtain block of to internet. to overcome these problems,
network addresses from InterNIC and can itself classful addressing is replaced with Classless
assign address space as necessary addressing. as the name of Addressing schemes
2: Host Number: It identifies a host on a implies, the address are not divided into classes,
network and assigned by the local network however they are divided into blocks and the
administrator size of blocks varies according to the size of
TYPES OF IP ADDRESSING entity to which the addresses are to be
1: Classful Addressing: ln the classful addressing allocated. IPv6 is classless addressing
system all the IP addresses that are available are The Internet authorities have enforced certain
divided into the five classes A.B,C,D and E. in limitations on classing address blocks to make
which class A,B and C address are frequently the handling of addresses easier. These
used because class D is for Multicast and is limitations are as follows:
rarely used and class E is reserved and is not i) The addresses of a block must be contiguous.
currently used. Each of the IP address belongs ii) Each block must have a power of 2(that is
to a particular class that's why they are classful 1,2,4,8...)number of addresses.
addresses iii) The first address in a block must be evenly
IP Address Format divisible by the total number of addresses in
The 32 bits IP address is divided into four octets that block.
and each octet is written in eight bit decimal
numbers. These four octet are separated by
dots and ranges from O to 255
The binary weights of each bit in the octet is
128,64,32,16,8,4,2,1. The format of the 32-bit IP
address is illustrated in the figure
SUBNETTING its best to get a transmission through to its
is a unique and powerful feature that is destination. but with no guarantees.
exclusive to the TCP/IP protocol and is one of IPv4 Datagram Format
the reasons TCP/IP offers great scalability.
Subnetting allows network address to be further
divided, apart from the already established
classful boundaries, into smaller, more
manageable networks. This division provides for
unparalleled scalability and hierarchy, and gives
a network administrator benefits such as
reduced network traffic, less susceptibility to
broadcast traffic, network optimisation. and
greater ease of management. For example, if
you were to borrow one bit from the host
portion of a Class B network, your subnet mask Limitation of IPv4
would be 255.255.128.0 1: The IP address relies on network layer
Subnet Mask address to identify end-points on networks, and
There are two parts to the IP address, the each networked device has a unique IP address
network portion and the host portion. Node 2: Uses a 32-bit addressing scheme, which gives
assigned that IP address as well as other nodes it 4 billion possible addresses
that must communicate with it have no idea of 3: If a network has slightly more number of host
the location of the line between host and than a particular class, then it needs either two
network portions of the address. The subnet IP addresses of that class or the next class of IP
mask provides the answer to this dilemma. The address
subnet mask follows the IP address and details 4: Identified limitations of the IPv4 protocol are
the line indicating where the network portion of Complex host and router configuration, non-
the address ends and the host portion begin. hierarchical addressing. difficulty in
Like the IP address, the subnet mask is in a 4- renumbering address large routing tables, non-
octet, 32-byte format. trivial implementations in providing security.
An example of a subnet mask is 255.0.0.0. a QoS (Quality of Services) mobility and multi-
value of 255 means match all. Each of the three homing, multicasting etc.
configurable IP address
Classes has a default subnet mask:
1: Class A 255.0.0.0
2: Class B 255.255.0.0
3: Class C 255.255.255.0
IPv4 (INTERNET PROTOCOL VERSION 4)

IPv4 is an unreliable and connectionless

datagram protocol a best-effort delivery service.
The term best-effort means that IPv4 provides
no error control or flow control (except for error
detection on the header). IPv4 assumes the
unreliability of the underlying layers and does