Building Network Topologies

Building Network Topologies

Contents
1 Introduction ................................................................................................................................................................1
2 Abstract ......................................................................................................................................................................1
3 Steps for Building a Network Topology.....................................................................................................................2
3.1 Determine your eventual goal.............................................................................................................................2
3.2 Assess your existing network .............................................................................................................................3
3.3 Determine if you need aggregation.....................................................................................................................3
3.4 Select a technique for creating the topology.......................................................................................................5
3.5 Create the topology.............................................................................................................................................7
4 Examples of Various Topologies Used for Different Studies.....................................................................................7
4.1 Topology for Client/Server Performance Analysis.............................................................................................8
4.2 Topology for Application Deployment...............................................................................................................8
4.3 Topology for Analysis of the Backbone .............................................................................................................9

1 Introduction
This paper is one in a series of papers documenting various methodologies in solving network problems using
simulation. The methodologies are well supported by the OPNET family of software. The objective of these papers
is to provide general guidance in solving problems. It is recommended that the user tune the approach to address
specific cases. In the interest of maintaining focus on the methodology, product documentation is not provided in
detail though references may be made to some of the product features. Complete product documentation can be
found in the manual set or the online documentation that comes with the software.

2 Abstract
The initial step towards performance analysis of a network is to baseline the network topology. The method for
building the network topology varies depending on what information is to be obtained from the performance
analysis. For example, a study involving application deployment may require the entire topology explicitly modeled.
A backbone analysis may require the backbone portion to be modeled in detail but allow the remaining portions to
be abstracted, a client-server performance analysis may require modeling of a single path between the client and the
server.

The software supports many different techniques for building network topologies. Topologies can be created
manually or imported from tools that perform auto-discovery. The model library that comes with the software
provides models of various devices used in current day networks. Devices that aggregate sections of the network
(such as LAN segments, ATM/Frame Relay clouds) are provided in order to simplify the network topology and
improve simulation performance.

This paper describes using the software to do the following:

• build complete network topologies

• build partial topologies

• model a single path between two devices

Confidential Information: Do Not Disclose, Forward, Distribute, Page 1 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

Listed below are recommended steps that a user should follow during the process of building a network topology. At
the end of this section, case studies are provided as examples of how these procedures have been put to use.

Disclaimer: This paper contains statements based on the author’s experience using and testing the products.
Unless specified, statements related to products are those of the author and not facts or information directly from a
third party. Therefore, the statements and results in this paper are subject to risks and uncertainties. The following
are some examples of risks and uncertainties that may cause results to differ: integration of other products,
application to different networks (e.g., network size), changes in the products, and/or accuracy of imported data or
data files. Please note that despite the time expended on editing and verifying the information and related graphics
in this paper, it is possible that the information and graphics include technical inaccuracies or typographical errors.

3 Steps for Building a Network Topology

There are five basic steps outlined for building a network topology.

Determine Assess your Determine if Select a Create your

your eventual existing you need building network
goal network aggregation technique topology

3.1 Determine your eventual goal

The choice of devices for constructing a network topology and the method is very much a function of the goal
of a simulation study. Depending upon the context in which the software is being used, you may represent the
topology as a single path between two devices, a partial topology with portions of the network abstracted or a
complete topology with every device explicitly modeled.

3.1.1 Single Path

In this case, only the infrastructure supporting traffic between two devices of interest is represented. For
example, if the objective of using the software is to analyze a client-server application, it is only necessary
to identify the devices and networks constituting the single path between the client and the server. The
effect of the remaining portions of the network will be taken into account by representing the traffic that
crosses, and therefore affects, the path of interest. The advantage of this topology is that it enhances the
focus of the study to the main objects of interest, i.e. the client and the server. Since other devices are not
explicitly modeled, the simulation is very efficient. There is no loss of accuracy since the effect of the
remaining network on the client-server traffic is taken into account. You can easily and quickly perform a
number of what-if scenarios by changing client/server parameters or the amount of traffic in the
intermediate sections of the network.

3.1.2 Partial Topology

In certain situations, it may be important to represent portions of the network in detail while abstracting
other sections. For example, if the objective is to study the utilization of a backbone, the backbone portion
must be completely represented. Other sections of the network may be abstracted. However their effects
should be captured when traffic information is entered.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 2 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

3.1.3 Complete Topology

A complete topology is necessary when a particular problem is scaled across the network and it is
important to identify the impact of the problem on all devices involved. For example, if a new application
is being deployed and the objective of the simulation is to determine its impact on intermediate network
devices, it may be important to model the complete topology. Complete topologies may also be used when
the objective of the study is to move a server to various prospective locations and study the impact of each
choice on network resources (say link utilization).

Listed below are some examples of goals and recommended topologies:

Goals of the simulation Recommended Topology

Performance of a database application Single Path

Performance of FTP with different TCP settings Single Path
Performance of applications for remote users Single Path
Performance of a new application in the network Single Path
Effect of a new application deployment on the network Complete
Backbone analysis Partial
Moving from shared media to a switched network Partial
Trading off Fast Ethernet Vs Gigabit Ethernet Partial

3.2 Assess your existing network

Before beginning construction the final topology, it is important to have a complete picture of the existing
network in the form of a drawing or a map. It is important to identify all existing devices, their role, and the
protocols that are running in the existing network. It is also important to identify the flows and traffic patterns in
your existing network. Consider the following questions:

1. Is the network flat (predominantly switched) or segmented (routed)?

2. Where are the main servers located (WWW server, database server etc.)?

3. What is the main traffic flow (users accessing the database server, traffic through a firewall etc.)?

4. What are the sources of broadcast and multicast traffic?

3.3 Determine if you need aggregation

Before performing the final steps in generating a topology, it is important to determine if you can aggregate
portions of the network. Aggregation can be performed at the segment level (LAN segment) or at the subnet
level (IP subnet). Portions of the network, which are outside the corporate control such as the Internet or the
carriers, can also be represented as simple "cloud" objects with the appropriate latencies. Examples of the above
are illustrated graphically below:

Confidential Information: Do Not Disclose, Forward, Distribute, Page 3 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

A LAN segment can be aggregated into a shared LAN object. The number of workstations that the shared LAN
represents and the different applications generating traffic can be specified. The shared LAN automatically
scales the traffic based on the number of stations. Utilization measured on the shared LAN can be modeled as
background utilization on the shared LAN object.

Figure 1: Shared LAN Aggregation

Shared LAN segments connected to a switch may be aggregated into a switched LAN object. The number of
workstations that the entire switched LAN segment represents can be specified and would be set to the total
number of users on all the shared LANs that are utilizing the switch. Applications can be configured on the
switched LAN and the model automatically scales the total traffic based on the number of workstations.
Utilization measured on a switched segment can be input as background utilization on the switched LAN. The
switching speed for the LAN segment can also be specified.

Figure 2: Switched LAN Aggregation

The frame relay cloud can be used to represent a section of the core frame relay network. Typically, a company
uses a frame relay carrier service for long haul communications and the carrier infrastructure can be represented
as the frame relay cloud. Packet latency and discard probabilities can be modeled on the frame relay cloud.
Similarly, an ATM cloud can be used to represent core portions of an ATM network and an IP cloud can be
used to represent portions of the Internet.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 4 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

Figure 3: Frame Relay (or ATM) aggregation

Figure 2: IP Level Aggregation

3.4 Select a technique for creating the topology

Topologies can be created manually or automatically through a process of importing from tools that perform
auto-discovery. Topologies can be constructed automatically from router configuration files and text/XML files
as well. Each one of these techniques is discussed in detail below.

3.4.1 Direct Import

The software supports importing topologies directly from a number of vendor products. Each import
procedure varies slightly based on the information obtained from the vendor products. Refer to the product
documentation for details on the import procedure.

The VNE Server Environment

The OPNET VNE (Virtual Network Environment) Server product provides an on-line, continuously valid,
integrated view of your network. VNE Server collects network data from disparate sources, and
intelligently merges this information to create a unified network representation that can be used for network
planning, engineering, and operations. VNE Server can be configured for unattended operation,
automatically collecting data and maintaining the VNE Database, and providing continual status updates
about data management activities and exceptions. Designed for openness, VNE Server is a user-extensible
solution that can encompass virtually any data source. VNE Server provides a complete, architected
approach for managing network information.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 5 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

Vendor Products

You can import from vendor products such as HP Network Node Manager or Tivoli Netview. Refer to
the product documentation for the complete list of supported vendor products and procedures on how to
import from these products.

Text/XML Files

OPNET supports text-based or XML-based topology import. These text/XML files have a specific format
that can be found in the product documentation. Geographic location information can be supplied in these
files.

Router Configurations

OPNET supports import of router configuration files. Geographic location information can be supplied
with the router configurations. When you use router configuration files, the topology that is created will
have attributes that control routing behavior specified based on the contents of the router configuration
files.

The key features of the import process are listed below:

• The import preserves the network layout and hierarchy. The relative positioning of objects is
preserved. If objects are within subnets, the software will create subnets and place objects within
them.

• Devices are mapped accurately to the model library. The software maintains a large database of
device models (e.g. routers, switches, servers etc.) and their characteristics. During import, devices
are identified based on their function and vendor.

• Import provides aggregation. The topology can be imported with LAN level aggregation, IP
segment level aggregation or no aggregation at all.

• A Question/Answer database provides a method for dealing with unmanaged devices in case of
import from HP Network Node Manager.

• Import can repair structural defects in the network.

Before performing an import from any tool, it is important to view the topology in the network
management tool and ensure that it is an accurate representation of your network. Tools such as HP NNM
require devices to be managed and configured with special MIBs (Management Information Bases) so that
they can identify devices appropriately. If the devices are properly identified in the supporting tool, they
will be imported directly into the software. If there are any devices that are not properly configured or
identified, this can result in the software generating questions for the ambiguous objects. The software will
log all questions into an import log or a question-answer database, which can be viewed by the user. The
database also provides suggestions for correcting any ambiguity in the network. Once the questions have
been answered, these answers are used for subsequent imports.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 6 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

3.4.2 Manual Construction

Manual construction can be used when the topology is simple (in terms of number of objects and
complexity of interconnection) or when supporting tools that provide topology information for direct
import are not available.

To build topologies by hand, the software provides a number of object palettes that contain common
network devices and links used for interconnection. The software also provides a technique called "Rapid
Configuration" that allows you to quickly create standard topologies (star, tree, bus, mesh, etc.) containing
many devices with a few clicks. Once a topology has been created by hand, the software provides features
that allow the user to select a large number of objects and apply attribute values in one operation.
Consistency checking is provided to ensure that links are accurate and the interconnected devices are
compatible.

3.5 Create the topology

Once the steps above have been completed, you can build the network topology. The link consistency check
must be executed to ensure that the topology is accurate and there are no disconnected links. When the topology
construction is complete, you can specify attribute values on devices and run simulations. Some features that are
useful for this purpose are listed below. Details about these features may be found in the software
documentation.

• Find node: Allows the user to locate any node by name across the network. LAN objects or any objects
that represent many devices can be configured with the names of individual members that they represent
(aliases). The find node utility also locates a device by its alias.
• Logical object selection: Objects can be selected based on their type or any of their attribute settings.
Selection sets can be retained for further selection.
• Applying changes to selected objects: Changes to attribute values for objects selected manually or using
the find/logical object selection command can be applied with a single click.
• Configuring attributes: Attributes can be configured to represent characteristics of the device. Protocol
parameters can also be tuned.
• Selecting advanced models that contain many attributes: Models are divided into a three-tier structure:
advanced, intermediate and final. Advanced models contain numerous attributes that can be tuned based on
the different protocols they contain. Intermediate models contain a subset of the advanced model attributes
and Final models contain only very basic attributes that have to be configured by the user.
Once the process of building a topology is complete, you may proceed to the documents on modeling network traffic
and characterizing applications.

4 Examples of Various Topologies Used for Different Studies

Depicted below is an example network and corresponding topologies that have been selected for various studies.
The example topology selected consists of users on shared LAN segments. LAN segments on the same floor of an
office are connected to each other via a switch. All switches interface to a building router. A backbone spans various
campuses and connects the building routers together. Servers are located on a server farm in one of the buildings.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 7 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

L1
L8
Backbone
Switch1
Switch6
L2
Router1 Router3

Servers
Switch2 Switch5
L3 Router2

L7
L4
Switch3 Switch4

L5 L6

LAN Segments

Figure 3: Complete Topology

4.1 Topology for Client/Server Performance Analysis

If the network is being used to study the performance of an existing application in the presence of regular day-
to-day network traffic, a single path from the client to the server may be adequate. The effect of network traffic
is to cause additional delay for the application traffic. The network traffic is represented by background
utilization on the intermediate devices. Application traffic is modeled explicitly. This is an example of a hybrid
simulation. Hybrid simulation explicitly models the application of interest. The network traffic effects are
analytically represented to obtain both accuracy and simulation efficiency.

Figure 4: Single Path Topology

4.2 Topology for Application Deployment

If the network is used to study application deployment, it is important to represent all the traffic flows from all
clients to the servers. Since the objective is to study application response time and the effect of this application
on the switches and routers, it is not important to model in full detail of each LAN segment. LANs may be
aggregated into shared LAN objects with the appropriate number of workstations. Note that the LAN object will

Confidential Information: Do Not Disclose, Forward, Distribute, Page 8 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

automatically scale the traffic based on the number of workstations. The flows across the switches and routers
must be modeled explicitly as the effect of the new application on such devices is important in the study.

Figure 5: Partial Topology

4.3 Topology for Analysis of the Backbone

If the above network is used to study utilization on the backbone, it is important to model flows across the
backbone accurately. Modeling individual LAN segments and switches is not important. The shared LAN
segments along with the switches can be aggregated into switched LAN segments that generate cross traffic via
the backbone.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 9 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part
 Building Network Topologies

Figure 6: Partial Topology

CONFIDENTIAL INFORMATION
DO NOT DISCLOSE, FORWARD, DISTRIBUTE, SHARE, OR MAKE COPIES OF THIS DOCUMENT IN WHOLE
OR IN PART. This document contains confidential information and may contain information that is proprietary,
privileged, and/or exempt from disclosure under applicable law. This document is intended for the exclusive use of
the person to whom it is disclosed. If you are an unauthorized person, you are hereby notified that any viewing,
copying, disclosure or distribution of this information may be subject to legal action. All unauthorized persons must
immediately destroy the original documentation without making any copies or further unauthorized disclosure.

Confidential Information: Do Not Disclose, Forward, Distribute, Page 10 07-Feb-03

Share, Or Make Copies Of This Document In Whole Or In Part