White Paper: An introduction to network testing

An overlooked yet essential tool for achieving a thorough appreciation of a network’s bottlenecks is to undertake network testing

An overlooked yet essential tool for achieving a thorough appreciation of a network’s bottlenecks is to undertake network testing

Why network testing?

All network managers must determine the operational guidelines that their network services and users will find acceptable. These operational conditions can be considered normal for that network. Studies have shown that network managers are betting on increased bandwidth as the solution to network problems.

A well-defined and properly implemented network testing strategy will help the network manager predict the operation of the network.

When do we need network testing?

Comprehensive network testing will enable a network manager to maintain the network actively. This insight may allow the network manager to predict network operation under a given load, or anticipate problems created by new services.

What needs to be tested?

Some important network characteristics include:

Utilisation levels

Number of users

Application utilisation.

Each network should be evaluated individually. The network testing will then be developed around the relevant criteria for that network.

Network testing: understanding the past to predict the future

At trade shows, most vendors display the latest in high-speed communication technologies. The technology displayed is a mutation of legacy topologies and protocols. In most cases, demonstrations consist of an individual accessing a video from a server while performing a number of tasks (email and file transfer for instance). The displays focus on "multimedia" capabilities; regardless of the technology, be it Ethernet, ATM to the desktop, workgroup or backbone, everyone is promoting "multimedia."

Most network managers consistently find themselves working on or involved with user complaints on a regular basis. These complaints include poor access times, no access to resources or session dropouts. A rather interesting recent study indicated that respondents from a cross section of Fortune 500 businesses reported the following:

Approximately 50 per cent of network managers will move to new high-speed technologies within 6 months

Approximately 40 per cent will make the transition to high-speed technology within two years

Some 10 per cent of the respondents thought their networks were fine as they are.

Of even greater interest, 60 per cent of the respondents attributed the need for increased bandwidth to an increase in client/server activity. Only 32 per cent implied that multimedia applications were causing problems. Surprisingly, video-conferencing raised few bandwidth-congestion concerns. The study went on to provide a very effective and insightful comparison between a number of new high-speed technologies and communication techniques.

The study however didn't answer the following questions:

Do you know how many users you have on your network today?

Where is your greatest network expense incurred?

What are the peak, operational, and average utilisation levels on your backbone network?

Which applications tax your network the most?

In essence, how many network managers really understand the most fundamental operation of their networks? Network managers are expected to plan for the future. Unfortunately they must live in the present, which also means they must deal with the past. Due to the inherited nature of networks, the network's past is probably one in which the current network manager was not involved.

One of the solutions to today's and tomorrow's network concerns is to build an effective history of the network. By understanding the past, you will be able to predict the future, to a certain degree. This feat can be accomplished by developing a network testing strategy.

Network testing: a definition

Network testing is the actual measurement and recording of a network's state of operation over a period of time. It involves recording the current state of network operation to serve as a basis for comparison or control.

What factors should be examined in network testing?

The first question to answer is "What are the normal operating conditions for my network?" All network managers must determine the operational guidelines within which their network services and users can operate efficiently. These operational conditions can be considered normal for that network. It has already been established (from the previously mentioned studies) that network managers are betting on increased bandwidth as the solution to network problems.

The concept of increasing your bandwidth raises a number of questions:

Am I increasing bandwidth so garbage on my network moves faster?

Am I judging how much bandwidth I need by the number of trouble calls I receive?

Have I evaluated my network and determined that the only solution is an increase in bandwidth?

After I have increased my bandwidth with new technology X,Y, or Z, how do I know if I have improved the network given that I have nothing to compare it to?

An understanding of the normal operating conditions of the network will be a valuable asset in a troubleshooting scenario. Furthermore, a well-defined and properly implemented network testing strategy will help the network manager predict the operation of the network. If you establish reliable network testing, the network's ability to support new applications and new users will not be a matter of guesswork.

When to network test?

You do not want to perform network testing while troubleshooting a problem. In a perfect world, you would network test each segment or ring for an extended period of time, say one full week. A network testing schedule would then be determined by analysing the results of the long-term network testing. From this initial long-term network test, trouble spots would be selected and focused on. A trouble spot might consist of excessively high utilisation or perhaps high levels of error conditions. You will have to determine what is normal, acceptable, and unacceptable for your network. Considering time and resources, you may be unable to network test every segment. In this case, you should select your critical or problematic networks for network testing.

Comprehensive network testing will enable a network manager to maintain the network actively. A properly implemented network testing schedule will provide a valuable insight to trends or changes in the network's daily operation. This insight may allow the network manager to predict network operation under a given load, or anticipate problems created by new services.

What do we test in a network?

One of the first things to establish is what will be tested. Some important network characteristics are:

Utilisation levels

Number of users

Number of operational protocols

Error statistics

Application utilisation.

Each network should be evaluated individually. Network testing will then be developed around the relevant criteria for that network. Once you establish the network testing criteria, a standard format should be developed to allow you to record easily the network testing data. Most PCs come equipped with database tools that enable the user to develop a network-specific testing table.

The next step is to establish a central database. This is where all network testing and troubleshooting reports will be kept. This information will serve as a valuable reference for the network. The more information that is developed and maintained on each user, segment, peripheral, switch, bridge and router, the easier the task of network maintenance becomes.

At this point, the network testing begins. Inventory is the first thing on the list: establish a comprehensive list of hardware components. One of the rudimentary steps in an inventory should be to record the total number of components in the network. Wherever possible, you should access configuration files, print them out, and attach them to a form that includes the local vendor's name and telephone number as well as any Technical Assistance Centre (TAC) numbers that may be available. While you are at it, add serial numbers, part numbers, and version information. This information should be readily available. TAC personnel are likely to require this information when you call.

The first major step is now taken in establishing effective network testing. Once you have begun network testing, you must stay the course. Undertaking a programme of network testing will be of no benefit if network testing is performed only after network modifications or problems. Remember that you are developing a history of the network - a history which will prove an effective tool if it is developed properly.

Compiled by: Paul Phillips

(c) Wandel and Goltermann

This was last published in March 2000

Read more on Business applications

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

Hello, i need to clarify the following:
Is it mandatory to have a NOC for a data center?
Would a data center (of any tier) be considered safely operational without considering a NOC within the same project at which usually all NW and Telecom components are monitored?
Is there any document / code involved?
Cancel

-ADS BY GOOGLE

SearchCIO

SearchSecurity

SearchNetworking

SearchDataCenter

  • How do I size a UPS unit?

    Your data center UPS sizing needs are dependent on a variety of factors. Develop configurations and determine the estimated UPS ...

  • How to enhance FTP server security

    If you still use FTP servers in your organization, use IP address whitelists, login restrictions and data encryption -- and just ...

  • 3 ways to approach cloud bursting

    With different cloud bursting techniques and tools from Amazon, Zerto, VMware and Oracle, admins can bolster cloud connections ...

SearchDataManagement

Close