evaluation of wide area network perfomance

1

Table of Contents

Abstract......................................................................................................................2

Introduction................................................................................................................3Local Area Network (LAN)......................................................................................................................................3Wide Area Network (WAN).....................................................................................................................................4Comparison between LAN and WAN..................................................................................................................4Transmission control protocol (TCP/IP)..............................................................................................................5File Transfer Protocol (FTP)....................................................................................................................................5Web server...................................................................................................................................................................... 5Network Topology....................................................................................................................................................... 6OPNET............................................................................................................................................................................. 6Open system Interconnection layer (OSI)............................................................................................................6

Objectives....................................................................................................................8

Literature Review.......................................................................................................9Congestion Control For Fair Bandwidth in LAN..............................................................................................9

Alternate method...................................................................................................................................................... 11Effective bandwidth utilization in WAN networks........................................................................................11Time Response Optimization On Local Area Networks..............................................................................12Literature review summary.....................................................................................................................................14

Methodology..............................................................................................................16Configuration with simulation results.................................................................................................................17

First Scenario............................................................................................................................................................ 17Second scenario........................................................................................................................................................ 22Third scenario........................................................................................................................................................... 27Summary Table......................................................................................................................................................... 32

Advanced scenario 1.................................................................................................................................................33Advanced scenario 2.................................................................................................................................................35Advanced Scenario 3................................................................................................................................................39

Discussion..................................................................................................................40First Scenario...............................................................................................................................................................40Second Scenario.........................................................................................................................................................41Third Scenario.............................................................................................................................................................43Channel Capacity.......................................................................................................................................................44

Conclusion.................................................................................................................47

Evaluation of WAN PERFOMANCE, MODERN COMMUNICATION SYSTEMS

August 8, 2012

Abstract

In this paper several ways of network configuration and determining system performance

are analyzed, a review on bandwidth optimization and time response utilization is done.

The paper discussion is divided into two sections, the first section discusses on how to

configure two local area networks where by each network consist of two computers

installed on it. The second section of this paper discusses on analyzing and evaluating

WAN performance on a small company LAN that consist of 20 user PCs, three sharing

printers, a local file and an email server. Simulation on each section is done and the

results are well analyzed and explained, a brief conclusion of the report is done and a

future enhancement to improve the system performance is stated and explained.

Evaluation of WAN PERFOMANCE, MODERN COMMUNICATION SYSTEMSAugust 8, 2012

3

Introduction

A Network system is configured using different setups of network systems. It is widely

known that a network system can be setup using two main network configurations, these

configurations are local area network (LAN) and wide are network (WAN). In

configuring a network system one should refer to hardware and software environment,

which builds a complete network system. In a network configuration the following things

should be considered, transmission control protocol (TCP), Internet protocol (IP), File

transfer protocol (FTP), webserver and the network topology. The entire above-

mentioned are described in short explanations below.

Local Area Network (LAN)

Local area network is a computer network, which interconnects different computers with

a limited or small area; LAN setups are usually done in areas such as schools, homes,

small offices, laboratories and other limited areas.

The technologies which plays apart in building local area networks are the twisted cables

used as Ethernet and Wi-Fi wireless technologies, few of the companies and

organizations uses the new advance of technology which requires optical fiber.

Local area is more useful for sharing resources purposes, this include sharing files,

printers, games or other many applications. A local area network can be expanded to

connect to other LANs, to the Internet or other WAN.

Local area networks are built using in less expensive hardware’s, such as hubs, network

cables, Ethernet cables and network adapters. LAN can also be configured to wireless

technologies, which makes it be Wireless local are network (WLAN) can be expanded to

other LAN hardware options. Internet connection sharing (ICS) supports controlled

access to LAN resources and its used to configure local area network; it is a specialized

operating system of Microsoft windows. Ethernet local area network is the most common


August 8, 2012

LAN, a local area network can be as small containing two computers to large LAN

comprising of many thousands computers divided into logical groups called subnets.

Wide Area Network (WAN)

Wide area network is one of the network configurations that cover a broad or wide area,

this refers to any networks that links through metropolitan regional or across national

boundaries. Large business organizations and government mostly use this network

configuration. They utilize WAN to share data among employees, buyers, and suppliers

from various geographical locations. The most common and popular WAN is the Internet

it self. Internet contains segments such as VPN-based extranets; these are also WANs in

themselves.

Further more many wide area networks are research networks that utilize leased lines.

Compared to LAN, WAN is more expensive to set up.

Comparison between LAN and WAN.

LAN WAN

Definition: LAN (Local Area Network) is a computer network covering a small geographic area, like a home, office, schools, or group of buildings.

WAN (Wide Area Network) is a computer network that covers a broad area (e.g., any network whose communications links cross metropolitan, regional, or national boundaries over a long distance

Speed: high speed(1000mbps) less speed(150mbps)

Data transfer rates: LANs have a high data transfer rate WANs have a lower data transfer rate as compared to LANs

Example: Network in an organization can be a LAN Internet is a good example of a WAN

Technology: Tend to use certain connectivity technologies, primarily Ethernet and Token Ring

WANs tend to use technology like MPLS, ATM, Frame Relay and X.25 for connectivity over the longer distances

Connection: one LAN can be connected to other LANs over any distance via telephone lines and radio

Computers connected to a wide-area network are often connected through public networks,


5

waves such as the telephone system. They can also be connected through leased lines or satellites

Components: layer 2 devices like switches, bridges. layer1 devices like hubs , repeaters

Layers 3 devices Routers, Multi-layer Switches and Technology specific devices like ATM or Frame-relay Switches etc.

Fault Tolerance: LANs tend to have less problems associated with them, as there are a smaller amount of systems to deal with.

WANs tend to be less fault tolerant. as it consists of a large amount of systems there is a lower amount of fault tolerance.

Data Transmission Error:

Experiences fewer data transmission errors Experiences more data transmission errors as compared to LAN

Ownership: Typically owned, controlled, and managed by a single person or organization

WANs (like the Internet) are not owned by any one organization but rather exist under collective or distributed ownership and management over long distances

Set-up costs: If there is a need to set-up a couple of extra devices on the network, it is not very expensive to do that

In this case since networks in remote areas have to be connected hence the set-up costs are higher. However WANs using public networks can be setup very cheaply, just software (VPN etc)

Maintenance costs: Because it covers a relatively small geographical area, LAN is easier to maintain at relatively low costs.

Maintaining WAN is difficult because of its wider geographical coverage and higher maintenance costs.

Geographical Spread: Have a small geographical range and do not need any leased telecommunication lines

Have a large geographical range generally spreading across boundaries and need leased telecommunication lines

Table 1, comparison between LAN and WAN, http://www.diffen.com/

Transmission control protocol (TCP/IP)

These are the set of protocols or rules, which are used together with Internet protocol to

send certain data in form of data packets or message units between computers on a

particular network. In a given network configuration the Internet protocol takes care of

data delivery, while the TCP makes sure that data packets have successfully divided into

small units, which can be efficiently transmitted.


August 8, 2012

File Transfer Protocol (FTP)

File transfer protocol is a network-standardized protocol, which is used to transfer files

from one user to another user on a transmission control protocol (TCP) based network

such as the Internet. FTP is built and installed on a client server and uses different

controls and data connections between the client and the server.

Web server

Webserver can be referred to either hardware or software in a network configuration. It

acts as the hardware if it is a computer, roundly it acts as the software if it is the

computer application, and the main purpose of it is to deliver the web content that can be

accessed through the Internet.

Network Topology

Network topology is the topological structure of a particular network and maybe

represented physically o logically. A network topology contains some arrangements of

various elements of a computer; these elements can be links, nodes, workstations and

other configurations. As stated above a physical topology is the one which involve

arrangements of various network components such as cable installation, device location

etc. where as the logical topology is the phenomena which explains how data flows

within a network configuration.

OPNET

OPNET stands for Optimized Network Engineering Tool, It is an application, which is

used to analyze the performance of different network configurations, it provides

performance analysis for computer networks and applications. It is an important

application in networking; one can design, implement and test a network system using

OPNET. In this paper performance of a wide area network will be analyzed using


7

OPNET and different performance scenarios will be evaluated and explained according

to the simulation results provided by OPNET.

Open system Interconnection layer (OSI)

The OSI layer is very useful in network configuration; the models of OSI layer are used

in designing and troubleshooting any network, since the OSI layer is the standard base of

any network. The OSI layer is made up of seven models, which are, Physical, data link,

network, transport, session, presentation and application.

Physical layer

This is the lowest layer of the OSI model, it refers to transmission of data packets and un

structured raw beat stream over a physical medium, example wires, cables, radio waves

etc.

Data link

This refers to point-to-point or broadcast of frames through a transmission link with error

checking, hence provides error-free transfer of data frames.

Network

This layer controls the operation of the subnet, it involves in routing or forwarding of

packets through intermediate nodes called routers.

Transport

This refers to error free transmission of data; it transmits data from one end to the other

end with flow of control.

Session

This layer allows session establishment between running processes, it also handles

operation of users and coordination, it also recovers an interrupted connection


August 8, 2012

Presentations

This layer converts the information into accessible usable independent platform format

Application

This layer serves as the window for users and application services, it provides network

services into application

Objectives

To configure two local area network computers, consisting a minimum of two

computers each

To allocate static address for the computers in the network.

To perform wired data transfer between networks in the laboratory

To evaluate WAN performance using OPNET

To simulate its configuration and to show the results

To discuss on the response time for FTP and webpage downloads

To describe the advantages of having a redundant link

To discuss link utilization between LAN and the ISP

To provide future enhancement for the network to improve its performance


9

Literature Review

Congestion Control For Fair Bandwidth in LAN

In a simple knowledge of networking, it is genuine that high concentration of users in a

cell or a particular network leads to Internet drop offs and uncontrolled traffic at the

access point (AP). These problems lead to poor connectivity for users, link failure

resulting to performance degradation and a possible network collapse.

In this paper ways to overcome these problems are discussed and analyzed, through this

ways it is what is said to be congestion control. The method that will be used to control

congestion is based on the channel utilization. The aim of this method is to schedule

proportionally the traffic flow by backlogging the que length while meeting the quality of

service (QOS) requirements ( zibo shi, et al, 2009).

By using this method, link utilization will be improved and lead to allocation of fair

bandwidth to each user out of the total bandwidth available. The method uses link

utilization as the traffic/congestion indicator and notifies to the sender to adjust its data


August 8, 2012

transmission rate, by doing so congestion criteria is met. Note that “ the link utilization is

computed by the channel occupying time in a given time period”, (Zibo Shi, et al, 2009).

Below is the simulation of the method explained above

Figure 2, FRA sending rate and TCP congestion window under the same RTT simulation

scenario


11

From the above diagram it is clearly observed that all the four users represented as FRA

can share the link bandwidth equally, simply because at the beginning of the simulation

the faman rapid amplifier (FRA) flow spread their traffic congestion rapidly and un

equally, in the next figure the simulation shows the congestion has been spitted and rise

until they reach a steady state value of 0.2 mbs as the link speed with a very negligible

fluctuation. This is because the adjustment of the FRA sending rate is based on the

maximum available bandwidth of the local area network instead of packet loss (Zibo Shi,

et al, 2009).

However the transfer control protocol (TCP) submerged in this system detects traffic

congestion from packet loss at the source node and makes a rapid fluctuation of data at

the congestion window.

Alternate method

In any network system there exist inherent conflict between fairness and throughput, the

conflict occurs in many network systems. Many researchers have studied these problems

in context of (single hop) local area networks, where by several clients exchange traffic

through access points (AP). As another alternate way of reducing congestion, researchers

propose the use of multi hop extension to wireless local area networks, where by the

client traffic is forwarded through a series of client links (Qunfeng Dong, et al, 2006).

Hence by using multi hops it helps to diverge the traffic to each user fairly, avoiding link

cell congestion, which results to link failure.

Effective bandwidth utilization in WAN networks


August 8, 2012

In order to increase and bridge the speed at the LAN-WAN interface many enterprises

install a number of policy based devices to enforce the administrative policies over the

WAN access ( Ashok Singh Sairam, et al, 2006).

In many local networks the most conventional techniques of bandwidth management is

highly concentrated only on the outgoing traffic from the local networks, simply because

that is the traffic which can be easily controlled.

However in reality and technical validity it is the incoming traffic that causes problems

such as bottlenecks to the networks. In this review several methods are branched out

which can help in bandwidth utilization. Some of the methods proposed are the use of

simple load balancing to optimize the use of WAN in a certain environment of scattered

homes. This method controls the incoming traffic by diverting the outgoing traffic from

some of the nodes to other links of the network. Through this method the incoming

bandwidth is measured and the thresholds are defined, then the traffic diversion takes

place to maintain the defined thresholds (Ashok Singh Sairam, et al, 2006)

Time Response Optimization On Local Area Networks

LAN as one of the network configuration has effects on time response if not configured

well, by definition time response, is the time taken by a particular network to respond.

Some networks respond fast and others delay; different technical issues can cause this

problem.

On this paper we discuss and review on how time response can be optimized on a local

area network for file transfer protocol (FTP) and webpage downloads. Different methods

can be done to optimize time response on LAN, a suggested method on this review is

through a plan of active experiment. Implicit faults and bottlenecks are network faults

which have a higher influence on response time of a particular server, considering these

faults, the proposed procedure can allow us to save time cost and increase mor validity of

obtained diagnosis (A. Babich et al., 2008).


13

The active experiment procedure is done in the presence of the test load which provides

different variance of factors, this method of active experiment planning is mainly

directed to fault detection and is helpful for detecting implicit faults and bottlenecks

which are very difficult to locate on a fully configured system (A. Babich et al., 2008).

Further more researchers have found that most LAN installed in educational institutions

are not performing well in response time, and this leads to poor performance of the full

system, this lack of performance is due to the fact that little attention is given on

educational LAN and there work load is high since there purpose mostly is to control the

distribution of software. In this paper the characteristics of this class of network is

reviewed since response time is taken as the criteria of judging the system performance.

The system is tested using four different policies where several tasks are allocated among

servers and system response is investigated through a combination of a closed-form

queuing theory, and discrete simulation( Miles Kennedy and Aarti badami,2001). Load-

leveling is the primary method for testing all the allocation schemes.

Below is the discussion on how response time can be improved using the four different

allocation policies, considering economic basis.

The first scheme is, workstation specific allocation (WSA).

In this scheme, each server is said to be redundantly storing the entire software library

and designating a set of subsets to all the available nodes. In this scheme, load leveling is

achieved by assigning a number of work stations to each server ( Miles Kennedy and

Aarti badami,2001).

The second scheme is mainly based on package allocation thus called, Package Specific

Allocation(PSA). In this scheme all available file servers has assigned to itself a unique

subset of the available software library. In this scheme load leveling is satisfied by

choosing the subsets and mixing the software to all servers, by doing so the system is


August 8, 2012

likely to attract almost an even share of the total work load( Miles Kennedy and Aarti

badami,2001).

The third scheme is Package Usage Allocation(PUS)

This scheme uses the technique of package arrangement, packages are arranged in

descending order according to their popularity ( Miles Kennedy and Aarti badami,2001).

So by working down on the list in descending order, packages gets assigned to the file

server until the total sum of the mainline storage reach an approximate value of eighty

percent of the total cache memory ( Miles Kennedy and Aarti badami,2001). In order to

avoid trashing seldom used help files and some needed files, the remaining percent of the

cache memory is used as the safety margin, incase of addition of materials in the cache

memory, the servers use the longest un used discipline for selecting what files can be

over written with the added files.

The last scheme is state allocation scheme (SAS).

This scheme functions when a request is made at the file server, in this situation the

scheme looks for a file server if there is a possibility having the requested package in

cache, if the result is positive meaning that a server is available, then the work station

will be assigned to the server, if not the program reviews the workload on each server

and the workstation gets assigned to the file server with the least load (Miles Kennedy

and Aarti badami,2001).

Literature review summary

Title Journals

description

Methods used Advantages Disadvantages


15

Congestion

control for fair

bandwidth in

LAN

The journals are

about how to

control traffic on

local area

networks.

Channel utilization, the aim of this

method is to schedule proportionally the

traffic flow by backlogging the que length

while meeting the quality of service

(QOS) requirements, link utilization is

used as congestion indicator to notify the

user to adjust its transmission rate.

The setup of this system is

simple to install not

complicated.

The method is not costful

interms of installation.

It targets normal

local area networks,

not so advisable for

busy networks

Another was is the use of multi hop

extension to wireless local area networks,

where by the client traffic is forwarded

through a series of client links, this will

lead to fairly divide the traffic to

individuals reducing congestion at the

link.

The system is flexible, can

be used in both systems

normal LAN and busy

LAN’s which requires high

use of softwares.

The method uses

multi hop

extensions; hence

the system is

complicated, as you

have to take into

consideration

several link nodes.

A bit more

expensive compared

to the first method.

Bandwidth

utilization in

wide area

network.

The journal

discusses on how

to improve speed

at the LAN-WAN

interface with

utilization of

bandwidth.

The method used is load balancing, This

method controls the incoming traffic by

diverting the outgoing traffic from some

of the nodes to other links of the network.

Through this method the incoming

bandwidth is measured and the thresholds

are defined, then the traffic diversion

takes place to maintain the defined

thresholds

It easily controls the

incoming traffic by diverting

the outgoing traffic form

different link nodes, hence

traffic can easily ne share

through different WAN

workstations.

The method is

highly concentrated

on incoming traffic

only, outgoing

traffic is somehow

neglected.

Time response

optimization on

LAN

The journal

discusses on how

to improve time

response on LAN

for FTP and

webpage

The method used is the plan of active

experiment, the method is done in the

presence of the test load which provides

different variance of factors, this method

of active experiment planning is mainly

directed to fault detection and is helpful

The advantage of this

method is the ability to

detect implicit faults and

bottle necks which are very

difficult to allocate in a

network system

It is time

consuming, since

requires active

experiment and the

presence of test

load, also


August 8, 2012

downloads. for detecting implicit faults and

bottlenecks which are very difficult to

locate on a fully configured system

complicated set up.

Time response

and link

utilization in

busy LAN such

as educational

LAN

The journal

discusses on four

network schemes

which can be used

in improving

response time on

busy LAN such as

educational LAN.

Workstation specific

allocation(WSA)

Package specific

allocation(PSA)

Package usage

allocation(PUS)

State allocation Scheme (SAS)

The methods states how

package and workstation can

be allocated to improve time

response on busy LAN’s.

It gives out different ways to

improve time response.

Table 1, Literature review summary

Methodology

In this part of the report, all the flow of the work done is explained step by step. As

mentioned in the objective and abstract, the purpose of this report is evaluating the

application performance across WAN. Evaluation and testing is done through a software-

based platform named OPNET.

As an overview testing will be done to a network system of a small company called delta

soft technologies, which consists of 20 user personal computers, sharing three printers

and a local file with an email server. The users are expected to run different applications,

which in include FTP, video streaming, file sharing, email and web browsing. Also users

run locally served applications such as printing, accessing the database and using the

intranet system. Through our application testing in OPNET, our goal is to study the

response time of various tasks, this will require to look at response time of FTP

downloads and webpage downloads, also link utilization between the LAN and the ISP

will be briefly analyzed all this will fall onto the first scenario of our testing.


17

After an evaluation of the first scenario, second scenario will involve adding a T1 link

between the LAN and the ISP to double the available capacity; this will be done by

splitting the LAN into two smaller switched segments and adding the T1 link between

them. Load balancing will ensure that there is an equal utilization between the T1 links.

The last and the third scenario will be evaluating the effect of failing one of the devices

and see the advantage of having the redundant link.

Explained above is the short overview of what is going to be evaluated during the testing

procedures of our desired system below. Testing will be evaluated using three scenarios.

Configuration with simulation results

First Scenario


August 8, 2012

Figure 0, Set up of WAN consisting of 20 user PCs

In this first scenario the evaluation of the network performance is done for a busy hour of

the day, therefore simulation time will be set to 1 hour. The evaluation will begin after

the email and file-sharing server are properly configured.

Below is the configuration of the network system, which is going to be evaluated.


19

Figure1, small company LAN with one switch over WAN.

As seen above the file sharing application and the email server have been successful

configured, so the next step is to evaluate the network performance by viewing web

application response, FTP download response Time and the WAN download link

utilization.

Refer to the images below after a successful simulation result.


August 8, 2012

Figure 2, WAN download link utilization

The above figure shows the download link utilization, from the figure we can see that the

simulation is processed for a hour since it takes the value 60 minutes, and by average the

download link reaches a value of 92% for a given time interval. This download link

utilization percent is high; hence it will limit the availability of bandwidth to potential

user applications


21

Figure 3, FTP download response

From the above figure the simulation is also run for 1-hour busy day, and from the graph

above the FTP download response time is very close to 2.5seconds, it averages around

2.2 seconds.


August 8, 2012

Figure 4, web application response

As for all, web application response is also run in an hour time interval and the web

application response as seen from above is close to 1.3 seconds at an average value of 1.2

seconds.

Above are the screen shots of the first done scenario a brief explanation will be done on

the discussion part, the next procedure is testing the second scenario and see the response

changes of the graphs.


23

Second scenario

In this second procedure, a redundant link will be added to the network system so as to

double the capacity, load balancing is going to be used, and this will ensure an evenly

distributed traffic between the two links. With an additional link it is expected all the

results to reduce, from the previous scenario.

Refer to the diagram below

Figure 5, additional T1 redundant link to the network system

As seen from the above the network system is divided into two smaller segments, where

by each segment is connected to a switch, the LAN is connected to the Internet with two

redundant links (T1). Load balancing between the two links is done using enhanced

interior gateway routing protocol (EIGRP).


August 8, 2012

ADDED LINK

The first simulation done on the configured system above is the WAN download link

utilization; the simulation is run for a busy hour of the day. Relevant results are shown

below.

Figure 6, WAN link utilization of the below link (T1).

As observed above, the link utilization of the lower link has reduced from 92% average,

to an average value of 55%, these are the results expected from this scenario.

The following simulation is the results of the link utilization of the upper link, the

average percentage value is expected to be nearer to 55%, as this will assure us that

traffic was evenly distributed, satisfying load balancing condition.


LEGEND:Blue: first scenarioRed: second scenario

25

Figure 7, WAN utilization of the upper link.

As expected the average percentage of the link utilization is around 40.8% to 49%, these

values are close to 55%, this shows that load balancing is done to ensure an evenly

distribution of traffic towards the link. Note that only one graph is represented here

simply because the upper link was not present in the first scenario.

The next simulation results is based on the response times, our expectation in these

results is that an additional link will increase the response time meaning that the system

will be fast and the values will drop down compared to the first scenario.


August 8, 2012

Refer to the results below

Figure 9, Web page application response of the second scenario

As observed above, one can see that the web page application response has dropped

down from a previous value of around 1.1seconds to a current value of 0.45seconds; this

is the significant improvement of the system.



27

Figure 10, FTP download response of the second scenario

Our expectation requirements are met by referring to the figure above, it is clearly shown

that the FTP download response has dropped down from its previous average value of

around 1.25seconds to the current value of 0.6seconds, this is a significant improvement

of the system.


August 8, 2012


Third scenario

In this scenario one router of the configured system of the second scenario is going to be

failed, and relevant response of the system are going to be shown on the simulation

results, short description of the results will be analyzed. Our expectation to this scenario

is that the application response times and the link utilizations will respectively increase.

Brief description of this scenario will be explained on the discussion part of this paper.

Refer to the figure below

Figure 11, Configured WAN system with a failed router


FAILED ROUTER

29

By observing the figure above, one can see that one of the router of the system is failed;

this is indicated by an X sign on the second router. This condition is tested to see if the

additional redundant link (T1) has any significance to the system and how the system

will perform and react at an absence of one router.

Figure 12, compared results of all scenarios for the lower link utilization

As observed from above one can see that link utilization of the third scenario is zero,

simply because the router containing the link is terminated.


August 8, 2012

LEGEND:Blue: first scenarioRed: second scenarioGreen: Third scenario

Figure 13, compared result of the second scenario and third scenario for the upper link

Both of the second and third scenario contains two switches, but the difference is the

third scenario the router is failed, in this simulation the lower link is tested and it is

observed that when the router is terminated the users can still access the interne but it

will require high utilization, one can see that the utilization of the third scenario has

reached an average value of 92% to 93%, while the second scenario remains the same

because no termination of the router.


LEGEND:

Blue: Second scenarioRed: Third scenario

31

Figure 14, compared result of all scenarios related to FTP downloads response


August 8, 2012


Observing the figure above one can see that the average FTP download response of the

third scenario has significantly increased compared to the rest of the scenarios, this

shows that when one of the router is failed it will increase the FTP download response

leading to a slower system.

Figure 15, compared result of all scenarios related to HTTP page response

As expected the webpage response of the third scenario has increased compared to the

rest of the scenarios, this shows that when one of the router of the system I failed it will

increase the time taken for application to respond, this will lead to a slower system.Evaluation of WAN PERFOMANCE, MODERN COMMUNICATION SYSTEMS

August 8, 2012


33

Summing up from the third scenario, which compares all scenarios, one can realize that

an additional redundant link (T1) plays a very big role in link utilization and speeding the

application response and FTP download response, hence the system will respond fast.

Summary Table

Available scenarios Description Advantages Disadvantages

Scenario1 WAN configuration

having one s8witch

with one T1 link, one

router

Simple to configure

not complicated.

Not so expensive

Requires high link

utilization, which reduces

availability of bandwidth,

takes long to respond for

FTP download and web

page application.

Scenario 2 WAN configuration

having two switches

with an addition

redundant link (T1),

two routers.

Improves utilization

Improve response

time, hence the

system responds

faster, because

response time is

decreased.

Requires many nodes,

complicated in

construction, more

expensive.

Scenario 3 WAN configuration

having two switches

with T1 links, but one

of the router is failed.

NONE Requires higher utilization,

limits availability of

bandwidth, takes longer

time to respond for FTP

download and web page

applications, hence results


August 8, 2012

to a slower system.

Table, comparison between the three scenarios

The simulation results provided above, are the results obtained for the three different

scenarios for evaluating the performance of the WAN, as seen above different scenario

provided different outputs and responses, further explanation and details will be done on

the discussion part.

The next stage of this paper will base on the evaluation of the WAN using the advanced

scenarios. Three advanced scenarios are going to be evaluated and simulation results will

be provided and described shortly. Refer to the scenarios below.

Advanced scenario 1

In this scenario we are going to evaluate what will happen when providing the data rate

to be the same in the WAN links. Firstly data rate will be set in scenario 1 so as an

achievement of web page response of 0.5seconds is achieved. So in order to achieve a

webpage response of 0.5 seconds, we will provide a data rate speed of 1967750Mbps;

refer to the simulation result below.


35

Figure 16, Advanced scenario 1, achieving webpage response of 0.5s in scenario 1


August 8, 2012

As seen from the above result, achievement of an average webpage response of

0.5seconds is full filled. The next step is to provide the same data rate used above to the

network configuration of scenario 2, hence the data rate will be provided to both upper

and lower link of the scenario 2. Below are the results obtained.

Figure 17, results achieved after providing same data rate speed to scenario 2.

From the above result we observe that even though the same data rate speed has been

used, there is an improvement in webpage response, the system now responds at an

average value of 0.45seconds, this means that the system responds faster than the one

used in scenario 1. The reason to this is because of the additional redundant link, which

improves the application response and the utilization cost of the system.


37

Advanced scenario 2

On the network configuration in scenario 2, there is a continuous streaming between the

music server and several users. This is defined using the traffic demand object, below is

the diagram representing the streaming process between users and the music server.

Figure 18, advanced scenario showing users streaming to the music server

As seen from above, there are seven different users who are streaming to the music

server, so the following step will be changing the traffic of each user streaming to the

server, this will be done by editing the traffic (packets/second) and traffic (bits/second).

An observation on application response and FTP download response will be made in

comparison to the second scenario where the traffic was constant not changed.


August 8, 2012

Refer to the simulation results below.

Figure 19, advanced scenario 2 in comparison with scenario 2 for FTP downloads response

Observing the figure above one can see that when streaming has started there is a high

intense of bandwidth used and the system responds very slowly compared to the normal

system when streaming is not done. This concludes that when users traffic is varying and

that there is a streaming utilization this will affect the system response time negatively.


LEGEND:

Blue: Second scenarioRed: Advanced scenario 2

39

Figure 20, advanced scenario 2 in comparison with

scenario 2 for HTTP page response

As observed from the FTP downloads response, here also the band width is very intense and the

HTTP page response is very slow compared to the normal scenario 2, this is because of un

equally distribution of traffic between the users, hence leads to a slow system response, load

balancing is not satisfied between the nodes.


August 8, 2012

LEGEND:


Figure 21, link utilization of advanced scenario 2 vs scenario 2

By referring to the simulation above, one can see that the link utilization for the advanced

scenario has reached 100%, this is very high and it will require very high cost to utilize

this system because of the high intense of bandwidth used and un equal traffic

distribution.


LEGEND:


41

Advanced Scenario 3

In this scenario, failing one of the WAN links of the scenario 3 does observation and

testing of the system. The obtained simulation result of this advanced scenario 3 will be

compared with the third scenario. Results will be evaluated in terms of link utilization,

FTP download response, and webpage response. Refer to the simulation results below.

Figure 22, Failed link vs failed router, FTP, webpage response and link utilization respectively

From this scenario we analyze the result of the failed link and the failed router, when the

router fails the ARP table of the network deletes the entry of the router table, so all the

available traffic is redirected to the other router connected to the WAN. Similarly one of

the WAN link is failed the network traffic request data from the network and router ARP

table which is present in the network machines, but since the router is not connected to

the WAN link this causes traffic delay. As seen from the graph there is high utilization

cost for this advanced scenario because the results is based on the upper link which all

traffic is directed to it, causing congestion at the channel capacity. On the other hand the

response times doesn’t have a big difference with the third scenario but it also does not

give good response, the system is still slow.


August 8, 2012

LEGEND:Blue: Third scenarioRed: Advanced scenario 3

Discussion

In this part of the report a brief explanation of the three scenarios described in the above

sections will be critically evaluated and analyzed. The evaluation will base on each

scenario stated above, the advantages of having a redundant link will be analyzed by the

help of an advanced scenario which will require setting the data rate speed to be the same

in each link. Discussion will be done with the aid of few simulation results for the

redundant link. Below is the brief explanation of the configured WAN relying in three

experimented scenarios.

First Scenario

In the fist scenario the network is configured to have 20 different user PCs in a LAN

connected to a single switch on one router with one link connected to the Internet access.

Three criteria were taken into consideration to analyze the network performance. These

criteria are WAN download link utilization, FTP download response, and webpage

application response (HTTP page response).

The simulation was successful done and valid results were obtained as how they were

expected. By evaluating the link utilization, the results shows that the average utilization

of this configured system is around 92%, this is very high for downloading hence it will

limit the availability of bandwidth for potential user applications.

FTP download response was recorded to be very close to an average value 2.5 seconds

and the application response time (HTTP page response) was recorded to be an average

value of around 1.3 seconds, the application response times are not so bad but they can

be further enhanced by adding an additional redundant link (T1), this will be seen in the

next scenario.


43

Second Scenario

The scenario involves adding of an additional link to the networks configured system;

hence an additional switch and additional router are implemented in this scenario,

discussion will base on the obtained simulation results described in the simulation part of

this paper. This is the most important scenario of all three scenarios; it is in this scenario

were advantages of having the additional T1 link (redundant link) are analyzed. In this

scenario results are expected to decrease in comparison with the first scenario, and both

of the links are expected to have a value, which is near to each other satisfying load-

balancing condition.

The simulation was successful done and valid results were obtained, evaluation was also

based in three criteria, which are download link utilization, FTP download response, and

application webpage response. The results were as follows the download link utilization

was recorded to have an average value of 55% percent for the lower link and 40.8% for

the new link, as expected the link utilization have dropped and two links are load

balanced to ensure that the traffic is evenly distributed between the two links.

FTP download response and webpage application response are also observed to have

dropped to 0.6 seconds and 0.45 seconds respectively. Therefore the additional redundant

link played a role in improvement of link utilization and response times for FTP and

webpage.

An advanced scenario can analyze the importance of an additional redundant link, in this

advanced scenario, we will analyze the first scenario and the second scenario ensuring

that the data rate speed is set to have the same value in both of the links of the first and

second scenario. In this simulation data rate is set to 1967750Mbps for all the links in

both scenarios.


August 8, 2012

Refer to the simulation results below.

Figure 16, webpage response having a data rate speed of 1967750mbps without and additional link.

Figure 17, webpage response having a data rate speed of 1967550mbs with an additional redundant link

The results above show that the webpage response of a data rate speed of 1967750mbps

is around an average value of 0.5seconds when no additional link is added to the system.

However when an additional redundant link is added to the system at the same data rate

speed, the system response decreases to an average value of 0.45 seconds, this shows that

additional of the redundant link improves response time of the webpage downloads, it

also improve link utilization and FTP download response as shown from the simulation

results described at the simulation section of this paper.


45

Third Scenario

The third and the last scenario is the other version of the second scenario, the scenario

configuration involves two switches, and two T1 links to access the Internet and two

routers but one of the router is failed. The expectation result of this condition is to have

higher utilization and higher time response, because only one router will be working to

operate a system with higher nodes. In this scenario evaluation will also be done in three

criteria, which are, download link utilization, FTP download response and webpage

application response.

The simulation is done and the valid expected results are successful achieved as shown

on the simulation section above in this paper.

The download link utilization in this scenario was recorded to have an average value of

around 93%, as expected failing one router will lead to high utilization this will cause

limited availability of the available bandwidth for the user personal applications.

FTP download response and webpage application response were recorded to have

average values of 1.5 seconds and 1.24 seconds respectively. These results are the

highest compared to all other scenarios, this shows that failing one of the router will lead

to a slower system, as the values of time response have significantly increased compared

to the rest of the scenarios.

To sum up, in comparison of the three scenarios discussed above, the second scenario

which requires an addition of a redundant link is a recommended network configuration

for soft technology company, as its results improves the response time of the system and

the download link utilization is successful improved.


August 8, 2012

Channel CapacityChannel capacity,

Channel capacity refers to the maximum rate of information in bits per second that can

be transmitted across a channel. Channel capacity plays a very big role in response time

of the systems; if the channel is over whelmed by traffic that it can’t handle the system

will respond very slowly or even can cause link failure.

Different transmission lines have different channel capacities, for a large LAN it is

advisable to use a transmission line which has a good channel capacity level. Channel

capacity can be used to calculate signal to noise ratio of unknown transmission lines.

Channel capacity is defined by a mathematical formula which relates three variables,

these variables include channel capacity (maximum bits per second), bandwidth and

signal to noise ration.

Below is the formula for channel capacity derived from Shannon Hartley theorem.

When observing channel capacity one should consider channel utilization. Channel

utilization is percentage of network capacity used by the network. It is given by the ratio

of the maximum rate of information transmitted at a given time by the maximum rate of

information that a given channel can handle multiplied by hundred percent.

For the network configurations on this paper the transmission line used is the T1 line

which holds the clear channel capacity of 1.544Mbits/second and the bandwidth value of

8Khz. By this default value we can be able to do calculations of the channel utilization

and to find the signal to noise ratio of our network system in the first two scenarios.


where by, C=channel capacityB=bandwidthS/N= signal to noise ratio

47

Firstly we will calculate the channel utilization of our first scenario, the channel capacity

is known which is 1.544Mbits/second, so we have to use the simulation result to know

the through put rate in bits per second, through put rate refer to the maximum rate of

information transmitted at a given time.

Refer to the simulation below.

Figure 18. Through put in bits/second for the first scenario

As seen from the figure above the average through put rate is around 1,400,000bits/

second. Now we can calculate the channel utilization.

Channel utilization= (max data rate transmitter at a given time/maximum data rate a

channel can handle)*100

Channel utilization=(1400000/1544000)*100= 90.7 %

From the calculation above we have seen that the channel capacity of the T1 link in our

system is utilized by 80.9% for the first scenario, this shows that the traffic against the

link is very high, so in order to solve this a new T1 link is to be added to double the

channel capacity. Below is the simulation result of the second scenario after a T1 link is

added, refer to the diagram below.


August 8, 2012

LOWER LINK

figure 20, through put in bits/second for the second scenario

As seen from the above one can observe that the traffic is distribute between the two

links, the upper link having a maximum rate of info transmitted at 600,000 bits/ second

and lower link the maximum rate of information transmitted is around 800,000 bits/s.

Below is the calculation of the channel utilization to see by how much is the channel

utilized after adding an additional link.

Channel utilization= (max data rate transmitter at a given time/maximum data rate a

channel can handle)*100

Channel utilization= [{(600,000+800,000)/2}/1,544,000]*100= 45.33%

The obtained result mathematically is around 45.33% this shows that the link has been

utilized approximately by 50%.

The results above are very similar to the simulated results of channel utilization on the

first scenario and the second scenario; this is the comparison between the simulated

results and the calculated result.

The following steps are finding the signal to Noise ration of the system. By default the

channel capacity of a T1 line is 1.544Mbit/s and the transmission line has a bandwidth of

274878 events per second or frames per second, from this result we can calculate the

SNR as shown below.

1544000=274878log to base 2(1+SNR)


UPPER LINK

49

SNR=48.08 or 16.8 DB. This result is the signal to noise ration of the first scenario; the

following calculation is based on the second scenario, which has a bandwidth of 390246

events per second.

1544000=390246log to base 2(1+SNR)

SNR=14.52 or 11.62 DB

As seen above one can see that the first scenario has a signal to noise ratio which is

higher that the second scenario.

Conclusion

To sum up all the objectives of the report are met, as everything stated on the objective is

done and well analyzed. In the first scenario evaluation of FTP, webpage response and

channel utilization was well analyzed and explained. The second scenario, which

involves adding of a redundant link, was critically evaluated and the advantages of the

redundant link were well stated and proved by the help of the simulation result. The

effect of failing one router was observed and it was analyzed in comparison with the

normal scenario where one router is working perfectly fine. The help of simulation

results showed the relevant results and were explained accordingly.

For further evaluation of the system three advanced scenarios were done and observation

of what differences they will create to the system were observed. By referring to the last

scenario WAN link was failed and the effects of it were compared by the second scenario

were one router was in function. The help of simulation results explained effects of both

conditions. Channel capacity is well analyzed as shown on the conclusion and manual

calculation is done to prove the relevance of channel utilization, the results obtained are

close enough to the ones obtained in the simulation result.


August 8, 2012

Future Enhancement

The response time is the critical element in the network. But since under some

circumstances the utilization is high then the response time gets lower as it can be

seen in Scenario1 & 2. Hence the network can be enhanced in future by active

planning of the peak hour traffic. The clients can be prioritized and be diverted on

an external link there by the reducing the effect of the lower response time on the

active clients in the network.

The second enhancement that can be considered in the above network for the

enhancement of the response time is the Ethernet connectivity technique that can

be employed in the network to boost the response. There common method used

here is direct MDI to MDI –x connectivity that can delay the response when the

client needs to connect proportionally instead there can be a cross over connectivity

between the clients that would enhance the response time by cross sectional

connecting each of it.

By Arnold Gabriel Ruhumbika

Site: http://www. inauzwa.com


http://www.inauzwa.com/

evaluation of wide area network perfomance

Documents

ashok singh

leased telecommunication

detecting

timemaximum

http page

file transfer

multi hop

ftp download