The use of VoIP in the corporate world is growing rapidly, driven by a combination of increasingly mature technology and a desire to reduce costs.

A single network infrastructure should enable organizations to reduce capital expenditure and create a more homogeneous environment that is easier to maintain, monitor and manage.

However, using the network to transport voice as well as data naturally reduces the amount of traffic it can support.

Moving to VoIP gives excellent results when executed properly, but requires careful planning and the right tools to avoid poor performance and reduced efficiency. If call quality is poor, users simply won't use it.

Planning a VoIP Implementation

There are two key points to consider when planning a VoIP implementation: the increasing capacity demands, and the nature of packetized voice traffic, which affects both voice quality and bandwidth use.

All packets are subject to latency, jitter and loss as they traverse the network. Data packets use TCP, which is connection oriented. If there is a delay, or receipt is not acknowledged, the protocol times out and data is resent, so events go unnoticed or have minimal impact.

In contrast, VoIP utilizes UDP, which is inherently connectionless. If a packet is lost, or delivery is taking too long, the sender has no mechanism to resend or adjust the rate at which data is sent.

Packet loss of more than 5% will start to affect voice quality. As a result latency, jitter and packet loss can have a devastating effect on call quality, rendering conversations unintelligible.

VoIP's inability to adjust for network conditions also means that it uses whatever bandwidth is available. TCP can and will adjust, so if VoIP is using a large proportion of the bandwidth TCP traffic will see low availability and applications will slow down.

Adding a significant number of VoIP users can impact utilization of network segments, reducing both voice call quality and the speed of standard TCP applications.

Quality of Service (QoS)

To solve this and preserve the integrity of voice calls requires two different classes of service. Most networks use QoS technologies to protect and prioritize VoIP traffic by tagging it at the device level with a queue marker (a Differentiated Service Code Point or DSCP) and setting parameters for how devices in their network treat it.

It's usually allowed top priority in being forwarded through the device, as well as some type of rate limit to ensure data applications continue to perform at the levels users expect.

Engineers can also analyze management information to help them adjust the interaction of VoIP with the infrastructure.

They can identify issues such as a lack of bandwidth in specific sectors, determine whether applications such as file sharing or streaming media are impeding VoIP performance, and how traffic should be shaped to prioritize the most business critical applications.

  • Roger Holder is Fluke Networks field marketing manager for network and application performance solutions for the enterprise. He has 20 years' experience helping organizations design and manage the performance of mission critical business applications.