IMPLEMENTATION AND ANALYSIS OF A
WIDE AREA NETWOK

(A FEASIBILITY REPORT)

BY
ISAIAH ADEBAYO
STUDENT’S NAME

WITH
3121658
STUDENT’S NUMBER
COMPUTER SYSTEMS AND NETWORKING ENGINEERING (CSN)-FULL TIME

SUBMITTED TO: DR VINCENT SIYAU (SUPERVISOR)

TABLE OF CONTENTS
CHAPTER ONE
1.0 AIM.............................................................................................................................Page 3
1.1 OBJECTIVES.............................................................................................................Page 3
1.2 INTRODUCTION......................................................................................................Page 4
1.3 EQUIPMENTS USED............................................................................................... Page 4
CHAPTER TWO
2.0 METHODS AND PROCEDURES............................................................................Page 5
2.1 PRESENT NETWORK OVERVIEW........................................................................Page 5
2.2 MAIN OFFICE NETWORK......................................................................................Page 7
2.3 SUBNET OFFICE NETWORK.................................................................................Page 7
2.4 BUSY NETWORK SCENARIO...............................................................................Page 7
CHAPTHER THREE
3.0 PROPOSED NETWORK SCENARIO......................................................................Page 9
3.1 TECHNICAL ANALYSIS.........................................................................................Page 9
3.2 DISCUSSIONS.........................................................................................................Page 13
3.3 ESTIMATED COST.................................................................................................Page 13
CHAPTHER FOUR
4.0 CONCLUSION.........................................................................................................Page 14
4.1 RECOMMENDATION.............................................................................................Page14

ABSTRACT
Optimal networks don’t just happen. They require proper planning and execution while considering a variety of factors. Some of these factors are highlighted in this report. Network congestion is a major challenge in growing organisations. Some of its underlying causes have been analysed as well.
There are various factors responsible for creating optimal networks. One of which is ensuring that there are multiple servers dedicated to each application on the network. In the fictional scenario described in this report page response time for services such on Database query, email, web and HTTP applications have been analysed.

1.0 AIM
To show how network performance is affected by background traffic in a Wide Area Network
1.1 OBJECTIVES
To Design a fictional WAN and show its implementation
To show network performance without background traffic
To show network performance with background traffic
To compare the existing average throughput and response time for Data Base Query, HTTP, Email and FTP
To draw conclusions based on the results generated
1.2 INTRODUCTION
The utilization of a network is usually dependent on the background traffic and the number of active users. This report seeks to investigate the effect of background traffic on a Wide Area Network with regard to network related protocols.
Analysing network protocols and services requires subjecting the target system to realistic Internet conditions. It is expected that competing background traffic will influence service and protocol behavior. It is the goal of this study to investigate and study the extent of such instances.

A device load is usually defined as a percentage of total capacity on a server, LAN or other nodes. The background utilization specifies the load as a percentage of the total utilization on that node. Usually a baseline load is a static throughput on a link, node or connection. One of the key elements is to generate the realistic background traffic to stimulate the load in network caused by various traffic sources. The consideration of background traffic gives more accurate features of the actual load in the network [1].

It has been observed that a number of research works have been done in different environments with statistically defined constant execution time of each individual service and CPU utilization of LAN devices. [1]

This study investigates the impact of background traffic on certain protocols in a wide area network using an OPNET simulator considering various scenarios.

1.3 EQUIPMENTS USED
The following materials were used in the simulation of this network. Cisco Layer three Routers 10 BaseT_LAN Ethernet 16_Switch 100 Base T Link cable PPP_DS1 Link cable PPP_DS 3Link cable Workstations or personal PCs Ethernet servers

2.0 PRESENT NETWORK SCENARIO
The Design of the present fictional network was done within a simulated scenario using OPNET IT Guru Academic Edition version 9.1.
2.1 NETWORK OVERVIEW
The present network under study consists of five subnets located in Dublin, Glasgow, Manchester, Birmingham and London respectively (as shown in Fig 2.1 below). Fig 2.1 An overview of the Network
The Application and Profile Configurations are placed into the grid and their attributes are configured and set up (as shown in the Fig 2.2 and Fig 2.3 below). Fig 2.2 Profile Configuration set up Fig 2.3 Application Configuration set up Each subnet apart from the London office is made up of 10BaseT_LAN connected to one Cisco 4500 layer three Router via a 10Base T cable. (as shown in Fig 2.4 below) Fig 2.4 illustrating the set up in each subnet
For the LAN in each office subnet, the number of workstations was set to 100, and the four rows for the applications supported profiles were each configured (as shown in Fig 2.5). Fig 2.5 LAN supported Profiles settings set up

2.2 MAIN OFFICE NEWTORK
Four Ethernet servers are placed in the London office which serves as the Headquarters of the fictional company. Each of these servers are connected to an Ethernet 16 switch and the switch connected to a Cisco 4500 Router using a 100Base T cable (as shown in Fig 2.6 below). Fig 2.6 London Office Network
Here the Data Base server also serves as the main server that supports all the other applications namely: Email, FTP and web applications. This is grossly inadequate and would cause poor network performance due to network congestion.

2.3 OFFICE SUBNET NETWORK
The PPP_DS1 link is used to connect between the neighbouring subnet offices in the base line scenario. But the PPS_DS1

3.0 PROPOSED NETWORK SCENARIO In proposing a new network there are a number of issues to consider. These include 1. Estimated growth of the network over a given period of time
Networks must be designed to accommodate growth in such a way that upgrades can be easily done within the shortest possible period. It is therefore necessary to project the expected network growth as this would affect how soon the present scenario becomes obsolete 2. Estimated network traffic load
It is also important to determine the network traffic before embarking on the design of any network. Heavy networks require very reliable and robust network infrastructure than networks with less traffic 3. Estimated cost of maintenance
Cost is an important factor to bear in mind while designing a network. Trade off between cost and performance is a variable every network designer must bear in mind
Bearing these factors in mind the major change is made in the main office where Four Ethernet servers are placed in the London office which serves as the Headquarters of the fictional company as against the existing scenario where one server is made to support all the applications under analysis. Each of these servers are connected to an Ethernet 16 switch and the switch connected to a Cisco 4500 Router using a 100Base T cable (as shown in Fig 2.6 below).

Fig 3:1 Proposed network overview
An upgrade of WAN links will be considered as well as in the new network. Due to economic considerations, the choice of prospective WAN links may be restricted to DS3/T3 (45 Mbps) or equivalent lines as against the DS1/T1 lines used in the present network.
For testing network performance of both scenarios simultaneously statistics have been ran. The Data base query, Email download, FTP download and HTTP page response times are the chosen statistics for this study.

3.1 RESULTS ANALYSIS
The results generated after simulation was run are shown below. The response time average for DB query, Email, FTP and HTTP is illustrated in the graphs below for the network with and without background traffic Fig 3.3 DB Query time average graph Fig 3.4 Email time average graph Fig 3.5 FTP time average graph Fig 3.5 HTTP time average graph
From the above graphs we see that there is significant difference in the response time for each of the investigated protocols. From this result it is observed that the there is a clear difference between the proposed network and the present network arrangement.
The page responses time in all applications are greatly minimized because of the fact that each server is dedicated to a particular application as against the previous network where all applications share the same server. This is not good enough. This shows that the present network needs to be upgraded for better performance
3.2 NETWORK UTILIZATION USING DS 1 LINK The graph and table below shows the point to point utilization of the network using DS 1 link.

Fig 3.6 Utilization graph of background traffic using DS 1 Link

Tab 3.1 Table showing average utilization from point to point
From the above graph and table it is observed that the network utilization is very high at about 96% on the average indicating that the network is overloaded. This would cause serious network congestion and delay. Hence the network performance needs to be improved.
3.3 NETWORK UTILIZATION USING DS 3 LINK
To achieve a better network performance I have created a third scenario from the busy network but changed the DS1 link between the subnets to DS 3 link. The graph and table below shows point to point utilization under this scenario

Fig 3.7 utilization graph of background traffic using DS 3 Link

Tab 3.2 Table showing utilization from point to point
From the set of graphs obtained from results generated while analysing the response times for the busy network using DS 1 Link and DS 3 Link interchangeably, a drastic drop in the average utilization of the subnets within the network was observed. The utilization dropped from an average of 96% to below 2% after the DS1 Link was changed.
The results generated from the scenarios above show that the DS1 link performs poorly across all traffic service types once there is background traffic. However, with the DS3 Link, there is very high possibility for network optimization. The DS1 WAN link has a data rate of 1.544Mbps which is far less than that of the DS 3 WAN link which has a date rate of 44.736 Mbps.

Although the DS3 links are far more expensive than the DS1 lines, they are very suitable in building networks with very heavy traffic. However there needs to be a trade off between cost and network performance. The priority of the client would determine which factor is considered above the other.

Disaster Recovery Services Pricing | Service | 3 Year Total Cost | 5 Year Total Cost | Project Management and Implementation Consultation, IBM–P-Series, VMware Servers (5 servers), Windows Servers (158 servers), Testing (10 days year1, 5 days subsequent years, Comms Services – 2 x dedicated 10Mb, Comms Services – Internet ATOT/ATOD up to 10Mb, Security Services, Core Infrastructure Services, Work Area positions for technicians | £126,017 | £297,031 |
Fig 3.8 Costing example for the proposed network
The table above shows a quotation for WAN implementation given by by Sunguard.

4.0 CONCLUSION
The experimental results demonstrated that the channel bandwidth utilization has much effect on the Quality of Service of background traffic. In addition, heavy background traffic highly affects the Quality of Service due to the network congestion caused by very high throughputs and network delays.

4.1 RECOMMENDATION
For optimal network performance and high quality of service it is important that the network administrator has a proper understanding of the underlying factors that affect the network performance. From the report we can deduce that when the background load is too much for the available bandwidth the network utilization becomes very high which invariably slows down the network speed and performance.

REFERENCES
[1] Sang-Soo Yeo et al. (2012) “Computer Science and its Applications”
-------------------------------------------------
Top of Form |
Bottom of Form “IT Guru Academic Edition 9.1”, OPNET Technologies Inc. http://www.opnet.com/university_program/teaching_with_opnet/textbooks_and_materials/index.html Ya Bao “IP Based Communication Networks” Lab Report Manual, LSBU