As any business expands, so does the network underpinning its communication infrastructure. But constantly adding more sites, users and applications presents a significant challenge for IT departments.

Performance can suffer as network traffic increases, and organisations usually decide to add capacity. But throwing extra bandwidth at the problem is an expensive option, and is rarely guaranteed to remove the obstacles to the smooth flow of application transactions.

Before committing to buying in bigger data pipes, companies should take a long hard look at their networks to find out how their bandwidth is being used, and whether or not the infrastructure is properly designed for its purpose, according to Scott Nursten, managing director of network consultancy firm S2S.

‘Many of our customers, including large government agencies and financial service companies, do not understand their network architecture. Documentation and diagramming is out of date, so they are already dealing with the wrong picture. Once that picture is accurate, it is time to understand the applications and platforms deployed on the network,’ he says.

‘It might be a question of design rather than optimisation. Companies need to look at what the network is meant to do and what it is actually doing.’

Eric Siegel, senior analyst at Burton Group, agrees in part, but points out that it is the nature of the applications in use today that is driving the need for more effective network management and traffic monitoring tools. Heavy volumes of web-based traffic and latency-sensitive applications such as voice over IP (VoIP) are pushing networks to their limits.

‘With web services the flow of transaction traffic is becoming confused and enterprises cannot keep up with it. If they are not monitoring network bandwidth and traffic, they should be,’ he says.

‘In the past all we had was transactions and bulk data traffic. If it was delayed by a few minutes, nobody noticed, but none of that is in the network now.’

Nursten adds that the virtual applications widely used to support remote access and contact centre environments in networks that span different countries can also test system design. Routing problems can result in downtime that has a huge impact on the business.

‘Products such as those from Citrix are latency- and time-sensitive. You get a lot of delay even moving the mouse around, which can kill your session,’ he says.

In some cases, IT departments can improve things easily by harnessing integrated network monitoring and analysis features to learn more about traffic at the application level. But the larger, distributed networks, ironically those which are most likely to suffer problems, tend to have grown organically over many years. They may have a mixture of new and old routers at different sites running different operating systems, and they do not offer the same traffic classification detail.

‘If the network is built end to end on Cisco equipment, it is quite easy because the equipment supports remote interface and dropped packet monitoring, as well as link utilisation. It can also deliver great statistics using NetFlow management features,’ says Nursten.

Netflow is a feature of Cisco’s IOS router operating system that can collect a wide variety of information from network devices, such as the source and destination of data packets, which helps to build a detailed picture of traffic flow across those devices. Juniper Networks provides a similar feature for its routers called cflow, and router vendor Huawei has another version called NetStream.

Almost all routers and switches support the simple network management protocol (SNMP), and SNMP traps can be set up to deliver basic status reporting on network nodes, including client devices and specialist appliances such as firewalls. There are also a large number of third-party hardware and software products that can provide more comprehensive traffic analysis.

See next page for what the experts say about network management