ITEXPO begins in:   New Coverage :  Asterisk  |  Fax Software  |  SIP Phones  |  Small Cells

Feature Article
May 2003

Continuous VoIP Assessment:
The Smart Approach To Delivering IP Telephony


Is your customer�s network ready for VoIP? If you care about customer satisfaction, that�s a basic question you need to ask yourself before implementing your next VoIP system for a customer.

Why? Because estimates indicate that as many as 85 percent of today�s router-based enterprise data networks are not ready to support VoIP without modification. As we all know, while data networks are built to be tolerant to network problems such as delay, jitter, and packet loss, these same issues can quickly impact the quality of real-time applications such as voice and video � often making these applications unusable. In addition, existing customer firewalls are normally configured in a manner optimized for data traffic and can greatly reduce the performance of IP telephony traffic. And guess who gets blamed when voice quality is unacceptable � or when the VoIP system starts interfering with other business-critical applications? You do, of course.
Therefore it is crucial to test a customer�s network thoroughly prior to VoIP deployment � and to continue testing critical network parameters after deployment. One approach that manufacturers, systems integrators, resellers, and service providers can use to minimize problems in VoIP networks is called �Continuous VoIP Assessment,� a phrase that was coined by Shoreline Communications, a specialist in enterprise IP voice systems created specifically to run Voice over IP across multiple sites.
Not surprisingly, the goal of the Continuous VoIP Assessment process is to avoid unpleasant surprises when it comes to VoIP deployment and ongoing maintenance. Shoreline�s product development approach strongly leverages the concepts of remote management and the need for a life cycle view. Shoreline has even put together an educational campaign for its resellers and its clients that outlines the importance of ensuring initial and ongoing network readiness when deploying VoIP.
While Continuous VoIP Assessment is a constantly evolving process, there are a few key steps that every provider of VoIP systems should undertake in order to avoid unhappy customers.

Step One: Select the Right Tool
We all know that you can sell more VoIP systems when you can prove that they deliver a quick return on investment, so of course you don�t want to add major new upfront costs to the VoIP deployment process. That�s why it is crucial to invest in one tool that can uncover and help you fix what is really happening on your customers� networks � throughout the entire VoIP assessment lifecycle.
In addition, selecting a tool that is software-based enables much faster and more cost-effective network assessment than hardware-based measurement tools. Software-based tools allow users to assess and monitor networks from one central location, thus eliminating the need to travel (or ship) to many locations to plug in hardware for accurate testing results. This reduces manpower, time and travel requirements, meaning that networks can be assessed within hours, not days or weeks.
In addition, the tool should have fully automated report generation features, which eliminate the work and the inevitable errors involved in the manual creation of technical or executive reports. Reports that can be printed out as editable Microsoft Word documents are also a bonus, as they allow you to quickly and easily share the results of your testing directly with your clients, and to customize the report with your own logo and company overview for a very professional-looking and consistent assessment report.

Step Two: Test the Correct VoIP Network Parameters
Every VoIP call is supported by signaling protocols that handle call setup, teardown, and dial tone. To initiate a call, a VoIP end-location goes off hook and informs a gateway or a gatekeeper about the new call, obtains a dial tone and dials the destination number. Once a call is established, VoIP gear at the two end-locations of the conversation negotiates the parameters, such as the codec, that will be used during the conversation.
During the conversation, voice is sampled at a constant rate, and the sampled bits are packed in Real-Time Transport (RTP) packets. These packets are then sent over the IP network using User Diagram Protocol (UDP). It is critical to test each portion of the VoIP call process to adequately assess the network readiness for the new application � as well as to determine what VoIP capacity the network can handle.
Testing the signaling portion of a call requires measuring the time it takes to establish a new call, the rate at which new calls can be generated and the success rate of completed calls. Testing the packet transmission portion of the call requires measuring key parameters such as delay, jitter and loss that can affect call quality.
In VoIP networks, call packet delay can create many quality problems. In fact, a packet that arrives late might as well have not been transmitted at all. Some of the factors that lead to delay include the use of different codecs, each of which adds a different delay; the distance between callers; and differences in bandwidth size, network load and network architecture. The jitter buffer delay, which results from holding packets in a buffer for a specified amount of time before decompressing them in order to smooth the packet flow, can also contribute to quality problems. The balance between packet and jitter is a precarious one. Ideally, you want both jitter and delay to be minimized. Unfortunately, this is tricky, because increasing the jitter buffer reduces the jitter but does so at the expense of increasing the delay.
What�s the golden rule for delay? Typically, people can tolerate total end-to-end delays of no more than 150 milliseconds, which can be considered toll quality. Conversation becomes annoying with larger delays and impractical at delays in the range of 400 milliseconds.
Measuring packet loss is also critical. As discussed previously, VoIP is transmitted using RTP packets that are transferred across the relatively unreliable UDP layer. Codecs are designed to handle lost packets by inserting a packet � often the previous packet or some kind of white noise � instead of the lost packet. This is fine as long as the percentage is very small. However, when more than one percent of the packets during a VoIP conversation are lost, users will experience a noticeable degradation in voice quality. The pattern of lost packets is also critical, as packets that are lost in bursts degrade quality more than packets that are lost sporadically.

Step Three: Fix the Problems
All of the above network parameters � plus many more � play a role in determining the voice quality of a network, which typically is measured using a Mean Opinion Score (MOS), where a �4� score (on a scale of 1 to 5) is considered toll quality.
Although MOS provides an important, standardized method of assessing a network�s ability to support VoIP, it is critical that your testing system measures not only MOS but also the network parameters affecting the network � and how they can be fixed. A MOS value doesn�t tell you which particular �hops� in the voice path are causing problems. It does not tell you if you have a device with misconfigured quality of service (QoS) settings. And it doesn�t tell you if auto-negotiation between any particular set of network devices has created a duplex mismatch.
Being able to peer into the network to identify specific points of delay and/or loss along a particular network path is a vital first step in fixing problems when they occur. Fortunately, in many cases, fixing the problems in a network does not require major surgery � it can be as simple as changing a router setting or increasing bandwidth on a particular segment.

Step Four: Test Again
After initial testing reveals one or more network problems and these problems are corrected, the next step is to perform a second testing cycle to ensure those actions were in fact effective. Typically, more problems rear their ugly heads, leading to yet another round of testing.
In order to fully determine the impact that VoIP has on the network, engineers should simulate VoIP usage across an entire business cycle and at all times of the day to fully understand how VoIP is affected by ebb and flow in network usage as well as how VoIP affects other business critical applications.
For instance, the VoIP system may work just fine � until the accounting department starts running its Oracle application to deliver invoices late in the month. A good VoIP testing tool can simulate that scenario to see how both the Oracle application and VoIP calls are affected due to the increased load on the network. A great testing tool allows you to schedule the simulation test to run unattended during off-hours and can generate real native-protocol application transactions to real servers (e.g. the Oracle server). This controlled traffic would be generated concurrently with the generated IP telephony calls to determine the effect of the application traffic on VoIP traffic and voice quality, and vice versa.
Running �what if� scenarios are also helpful in determining the impact of different situations on quality, such as �what if� I use a different protocol or codec; �what if� my customer service department suddenly gets overwhelmed with calls; and �what if� I change my QoS/DiffServ controls � to name just a few.

Step Five: Determine Ongoing Testing Needs
The only constant in life is change, and this is very true for communications networks as well. Adding new applications, new locations, or increasing network usage can all negatively impact VoIP quality. By conducting ongoing testing, you can proactively assess that impact and catch network glitches before they turn into real problems. Eventually, such software could even be used by help desks to reduce the amount of time it takes to resolve trouble tickets � by allowing customer service technicians to open a browser and immediately see problems within a customer�s network.
Both pre-installation and ongoing testing can even serve as a revenue enhancer for your company. According to some industry estimates, income from pre-installation testing from just one customer could bring in over $400 per month and could save that reseller $800 per month for every truck roll that was eliminated thanks to an increased ability to spot and resolve network problems remotely.
And that�s not even taking into account the non-quantifiable gains that result from increased customer satisfaction. The bottom line for VoIP manufacturers, systems integrators, resellers, and service providers is that investing in the process of �Continuous VoIP Assessment� can deliver not only happier customers, but a rapid return on investment as well.

Arne Lepp is sales engineer for Viola Networks,a company that specializes in solutions for IP network performance assurance and assessment to enterprises, manufacturers, integrators, and service providers. For more information, visit www.violanetworks.com.

If you are interested in purchasing reprints of this article (in either print or HTML format), please visit Reprint Management Services online or contact a representative via e-mail at [email protected] or by phone at 800-290-5460.


[ Return To The January 2003 Table Of Contents ]

Today @ TMC
Upcoming Events
ITEXPO West 2012
October 2- 5, 2012
The Austin Convention Center
Austin, Texas
The World's Premier Managed Services and Cloud Computing Event
Click for Dates and Locations
Mobility Tech Conference & Expo
October 3- 5, 2012
The Austin Convention Center
Austin, Texas
Cloud Communications Summit
October 3- 5, 2012
The Austin Convention Center
Austin, Texas