Convergence is king

An increasing number of organisations are moving to a converged infrastructure for voice and data communications. The benefits are clear: a single infrastructure enables organisations to reduce capital expenditure and create a more homogeneous environment. This should make monitoring and management more straightforward, improving performance and reducing MTTR and hence increasing user satisfaction.

However, using the existing network to transport voice calls as well as data creates increasing challenges for IT teams, who need to deliver the call quality users expect while maintaining data speeds to ensure organisations obtain the required return on their investment in a converged infrastructure.

When implementing unified communications (UC) there are two key problem areas: the network can become clogged by bandwidth hungry applications and employees may struggle to adapt to the new environment created by changes to the network.

Limited bandwidth is already an issue in many networks, but businesses are constantly adding bandwidth hungry applications. When UC is introduced, more and more of these applications such as voice and video will be driven over the wireless infrastructure. 4G rollout will see more VoIP calls made from portable devices, as this is the only kind of communication offered. Most business WLANs will not be prepared for this increased load, which will also highlight wireless problems that may have been previously hidden. Organisations need to ensure they have sufficient capacity when planning a UC implementation.

The issue of employee behaviour is less obvious. Voice over IP (VoIP) is a very different experience from traditional business phone calls: there may be no handset; the user may find it difficult to hear the person at the other end of the call due to background noise; jitter, latency or packet loss may make the call incomprehensible or cause it to drop out completely. Hence a VoIP implementation is likely to create many more calls to the helpdesk as users struggle to achieve the same experience that they had with circuit switched calls. If users find VoIP too difficult to use or regularly experience problems, they will find alternative solutions, and the organisation will not benefit from its investment in UC.

Managing the difference between voice and data packets

To implement UC successfully, IT teams need to understand the different characteristics of packetised voice and packetised data traffic. 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. Voice utilises 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. As a result latency, any jitter and packet loss can have a devastating effect on call quality, making conversations unintelligible.

VoIP will also use whatever bandwidth is available, while TCP can and will adjust, so if VoIP is using a large proportion of the bandwidth, applications will slow down. The solution is to use QoS technologies to protect and prioritise voice traffic while setting some type of rate limit to ensure data applications continue to perform at the levels users expect and demand. Managing all of this successfully requires a management solution which provides end-to-end visibility into the performance of both voice and data traffic.

The difference between voice and data traffic becomes even more noticeable in a WLAN. When implementing VoIP in 802.11 wireless networks, collisions in the wireless link reduce capacity and increase latency and jitter. The standard has a built-in algorithm to reduce collisions and hence lessen these problems, but this creates an additional overhead on each packet which will reduce the bandwidth available. If WLAN bandwidth is already at its limits, it may not be possible to add VoIP without increasing bandwidth.

A five step implementation process

We recommend a five step process to implement UC successfully.

1) Assess the environment and ensure the infrastructure is VoIP ready by carrying out a complete assessment of the network, from LANs and WANs to desktops and phones. Baseline and monitor actual usage and performance to find out if the data network needs to be modified to handle voice calls. If bandwidth is already at its limits in certain parts of the network, for example, this may need to be increased. Priority will need to be given to real time applications – not only VoIP but also videoconferencing/streaming – and these will require careful QoS allocation.

2) Check all elements continue to communicate properly during the implementation. Establish a performance baseline by generating synthetic VoIP traffic to confirm quality before full deployment. As the implementation progresses new issues may arise, such as new sources of interference in the WLAN, which will need attention.

3) When deployment has been completed carry out a stress test, looking at both overall performance and performance between specific sites. This will enable any potential bottlenecks to be identified and preventive action taken before they have a significant impact on users.

4) Actively monitor VoIP performance from various points on the network, including the core, individual routers and WAN links. VoIP problems can have a myriad of causes, from physical problems on the local loop and over utilised ports to misconfigured class of service settings and high levels of jitter within the voice application. Addressing this requires a broad management and troubleshooting strategy for the converged network. Real-time and back in time perspectives are vital to proactively circumvent emerging situations resolve ongoing anomalies as rapidly as possible, identify and correct intermittent problems, and optimise performance.

5) Educate users about VoIP. As we discussed earlier, making a VoIP call is very different from a traditional phone call, and users need to know what to expect and what to do if they experience problems. Some unified communications systems have an indicator of call quality, but in our experience these can be misleading. What a user may think is poor quality of service could be a software fault or other issue.

To achieve user acceptance, VoIP must provide high quality voice services without compromising the performance of data applications. This requires time correlated performance management data across traffic makeup, application performance, and VoIP stream quality by site, with the granularity to examine individual calls. The best management solutions integrate diagnostic data from the wired and wireless aspects of the network in a single interface and can quickly identify whether the problem is linked to the WLAN or somewhere else in the network.

Since most IT staff work in silos and are specialised in one specific area, having a solution that is aware of all areas provides a collaboration point that will help IT organisations understand and resolve the problem faster, leading to better collaboration, quicker MTTR and a higher quality of end user experience.

In essence, preparation, planning and a performance management solution are key so that the user is satisfied with the new method of communication. Change often leads to scepticism, therefore clear rationale and planning is key for the user to have the same or better experience post-unification.