Getting in Front of VoIP Quality Metrics
The adoption of VoIP (Voice over Internet Protocol, or Internet calling) is steadily rising in enterprises. Cost benefits resulting from using a shared data network and free long distance calling are compelling for large global businesses. In addition, many companies are motivated by the ability to integrate several advanced communications capabilities with voice communication. But the availability and quality of VoIP calls can be inconsistent: user complaints about VoIP quality can be challenging for IT to adequately track and troubleshoot. Typical ways to manage VoIP networks include tracking subjective user experience and network metrics, but these techniques do little to predict network performance and call quality.
Communication is the lifeblood of any company. Whether using traditional wireline phones or VoIP phones, effective communication is at the very heart of most critical business functions. Consider the critical need to communicate within a high-functioning business team, around a successful sales transaction, to maintain an ongoing customer relationship, to deliver effective service, or conduct interactions with supply chain and business partners. VoIP offers multinational corporations the ability to have a cost-effective and consistent communications platform for employees, whether they are on-site, on-the-go, or stationed remotely.
VoIP can also enable other value-added communications applications, like image and video streaming, fax, and email management. As VoIP is increasingly deployed in enterprises, it’s essential for IT to develop a strategy to monitor and measure the service for quality and availability. The challenge with VoIP is that tracking performance and user experience is not always straightforward.
VoIP was introduced in the mid-1990s, and relies on data networks and the public Internet to transmit voice communication. The outgoing voice communication is formatted into data packets for transfer across the data network, using data transmission and compression technologies. The origination and conclusion of a VoIP call relies on a range of networks, switches, routers, software, and devices. For example, a hosted VoIP service might be accessed from a telephone handset on a corporate data network and be transmitted to a home Wi-Fi network over to a laptop. A different VoIP call might be conducted entirely on a single corporate data network. The type, status, and location of the different devices, networks, and technologies changes from call-to-call.
Added to this dynamic complexity, VoIP and voice communications are highly sensitive to bandwidth and network traffic issues. Unlike a normal data package, which can be reconstructed with little loss in translation, voice communications require all data packets to arrive in the order they were sent with little lag time between packets. The quality of the call must be excellent — and measured — at both the sending and receiving ends of the call.
For IT managers, the task of VoIP quality assurance never ends. IP networks change constantly. Some of these changes can be predicted, but others cannot. For example, every time a device is added or a component changes, it can affect VoIP performance. Compounding the challenge is the unknown impact of external networks and devices. Ongoing, comprehensive monitoring helps network managers find problems before their users experience them.
Adding to the challenges is that user expectations are very high. Traditional dedicated telephony networks delivered consistently good quality and reliability, so users want comparable VoIP performance. To ensure that the VoIP experience meets user expectations and business requirements, IT has several options for measuring and monitoring the quality and availability of VoIP calls. An ideal approach is a combination of several options: surveyed user experience, call data reporting, and predictive call monitoring.
VoIP Monitoring
VoIP quality and availability monitoring focuses on a few key metrics, including latency, jitter, loss, and MOS. Most user experience complaints center on these elements.
Latency is the time a packet takes (in milliseconds) to leave its originating point and arrive at its ultimate destination. Network delays — whether due to traffic, slow network links, or other causes — cause latency. Latency ultimately leads to echo: the Darth-Vader-like quality of some VoIP conversations. Latency can also cause garbled audio, silence, and other odd sounds.
Jitter is usually noticeable as a delay in the sound during a conversation, and it can be surprisingly disruptive to the communication process. Jitter occurs when packets of voice data are received at the wrong time. It is measured by the amount of lapsed time between the delivery of consecutive voice packets. Jitter can occur due to power surges, bandwidth congestion, or other irregularities in the system.
Packet loss occurs when a network connection becomes overloaded with data or traffic, or because of poor quality infrastructure wiring. Packet loss results in a conversation that echoes, similar to having a conversation in a big empty room.
MOS (Mean Opinion Score) reports. This provides a numerical indication about users’ perceived call quality. Since the traditional manual/human MOS tests are quite subjective, there are now software tools that carry out automated MOS testing. Although they lack the human touch, these tests consider some of the network dependency conditions that could influence voice quality.
Options
If an enterprise is using a hosted VoIP solution, many VoIP vendors will provide data sample reports, event and call detail records to affirm that the services they have committed to deliver meet agreed specifications. While these metrics and reports can provide a rear-view mirror perspective to evaluate ring and delay success, call complete status, and call quality for users, they do little to help corporate IT predict, pinpoint or troubleshoot problems if users report issues. In addition to tracking and reporting VoIP network performance, hosted VoIP vendors frequently poll users about their subjective experience of call quality and provide MOS reports.
To obtain a more objective view of VoIP call quality and availability, or for on-premises VoIP networks, some IT organizations use tailored data center instrumentation to analyze calls internally. These tools can monitor and analyze attributes like audio quality, echo, volume, harmonic crackling, and distortion.
Like hosted VoIP monitoring and reporting, these techniques provide a good historical "rear view mirror" view if problems occur, and can help pinpoint issues on the network. But they can’t anticipate problems before the user experiences a poor-quality call, nor can they help pinpoint issues that lie outside of the corporate network.
A newer, more aggressive monitoring approach can actually verify the availability of VoIP service for users and the quality of VoIP calls by placing and simulating calls. This approach uses pulse technology to automatically place calls between locations — whether on premises or external phones. When a call is placed, the VoIP monitoring service plays music, enabling the system to measure attributes of loss, latency, and jitter. The system can also score and generate a MOS report.
A VoIP monitoring service like NETSCOUT’s TruView™ Live can trigger a user-defined alert if there are quality or availability issues. It can configure the frequency, locations, and alert thresholds according to their specific requirements. This technique also offers proactive and consistent monitoring of the networks, and can show if a remote network is down, a hosted VoIP provider is experiencing traffic problems, pinpoint a cabling or device issue, or indicate if any of a number of common, or uncommon problems, is occurring.
Getting Ahead of VoIP Quality Issues
Imagine the consequences: a senior sales executive at a multinational corporation has scheduled a phone meeting and presentation to update the company’s biggest customer about the company’s product roadmap. As the executive launches into the discussion, her voice morphs into robotic clicking and whirring. As the customer tries to interject and ask for clarification, her cyborg-like comments override his questions. Somehow, they manage to determine that the call cannot proceed. The whole experience leaves a bad taste in everyone’s mouth. The meeting must be rescheduled to accommodate everyone’s busy calendar, the customer is left wondering whether attention to detail and quality are attributes that are held in high regard at the company, and the IT department is required to troubleshoot the problem after the fact with limited information to help them diagnose the issue.
As corporations increasingly deploy VoIP communications systems to enhance communications capabilities and save money, it will be essential for IT to determine best practices to ensure quality — before the fact — for both the big and little calls happening every day in every corporation. Ideally, the quality-monitoring process enables reporting and alerts, and should be proactive and continual, predictive, configurable, and provide full visibility to internal and external networks and devices.
Using a variety of VoIP monitoring techniques, IT organizations can pinpoint voice quality problems and identify the issues fast. As a result, the employees of organizations that use VoIP technology can be confident that an essential business tool will support their efforts and results.