As I travel the world teaching students about the world of contact centers, I am amazed by the growing number of students who are asking about Voice over IP (VoIP). Most are confused about how VoIP differs from a normal PBX when it comes to delivering calls and many of these students are wondering if the PBX as we know it today is the antique many vendors suggest that it is. I am also frequently asked about why the move to VoIP is not a stampede. The answer to these questions begins with an understanding of how calls are made using a PBX as compared to how calls are made using VoIP.
When a call is placed from a standard telephone to another standard telephone, the phone company dedicates a set of wires or circuits for the exclusive use of the call. This dedicated set of circuits is essentially an electronic model of 2 soup cans and a piece of string. The “string” in the case of a phone call is made up of a number of wires that are electronically connected together end-to-end by the phone company. Like the children’s can and string model, a modern phone call is between 2 phones or end points with the line or “circuit” being dedicated to the caller and the recipient for the life of the call. As soon as one of the parties on the call hang up the phone, all of the connected wires or circuits are released for use by other calls; it is like someone cutting the string between the soup cans. The key concept to grasp is that a phone call involves a dedicated connection between 2 points for the exclusive use of carrying voices from one end to the other and back.
VoIP operates using a different model. In the VoIP world, all sounds are converted from analog form, like those you hear through a standard telephone handset, into digital form, a stream of 1’s and 0’s. The stream of 1’s and 0’s is assembled into a series of small groups called packets. These packets travel across a variety of wires along with a packets from other devices until they reach their destination. Thus when something is spoken during a VoIP call, a digital stream of 1’s and 0’s is created which are broken up into a numbered series of packets. When that series of packets reaches the other end of the call, the digital stream is re-assembled and converted back to its native analog form and heard through the phone.
Thus you have the connection between the world of voice carriage and the world of data carriage. VoIP, in a nutshell, is the carrying of voice traffic over the traditional data processing hardware. Instead of voice switches electronically connecting lots of lines together to form a dedicated circuit, VoIP uses a series of data routers to move packets of data from one phone to another. It is the potential cost-saving opportunity to share the data network for both computer traffic and voice traffic that has brought about the interest in VoIP.
The advantages of sharing a single network for both voice and data are obvious:
• Why pay for voice lines and data lines that all go to the same places when 1 line will do?
• Why have 2 sets of staff, 1 for the voice equipment and 1 for the data equipment, when I can have a single staff?
• Why pay for long distance calls between offices when I can use my existing data network?
Those who understand only the basics of VoIP see these advantages as compelling reasons for an immediate shift from the PBX world to the pure IP world. Those who understand the true nature of VoIP know how to properly evaluate when VoIP makes sense.
The Dirty Little Secrets
A single network to carry both voice traffic and data traffic rather than a network for data and a separate network for voice is certainly compelling. Less staff, less complexity and less vendors appeals to most business people. Before anyone charges off to replace their existing PBX with a series of VoIP switches, it is important to understand some very important differences between the traditional data world and the traditional voice world. Knowledge of these differences is vital for accurately assessing the value of VoIP in any given situation. These differences can be categorized into 4 areas: line sharing, network expansion, capacity planning and prioritization.
Let’s start with line sharing.
The VoIP model is very similar to the old-fashioned party-line phone system found in the traditional voice world. In a party-line system, multiple people share a common phone line. When you want to place a call, you must first wait for the line to be available. If you pick up your phone to make a call and hear that someone is already on the line, you must wait until they are done before placing your call. When you have secured the line to place your call, the line is yours until you hang up the phone. As long as the line is in use by another party, everyone else who shares the line must wait before they can place their calls.
Fortunately, computers do not “talk” to each other in the same way people do. Computers have been designed to talk in short, quick bursts. Access the line, ship some data to another computer and release the line. Thousands of times per second, computers can get on and off a line. This pattern of short bursts allows many computers to share a single line without seriously impacting the other computers on the same line. Human conversations are not shorts bursts of words but exchanges of sentences, sometimes lengthy sentences. VoIP combines these opposite patterns together on the same wire and it comes with a cost.
Computers have been designed to wait patiently for the line to become available. In the voice world, waiting to hear what the other person has said is not an acceptable condition. We have been conditioned to expect that when talking into a phone, the party on the other end hears what is said as soon as it is spoken. Telephones are not walkie-talkies; we don’t accept delays on the telephone. The need for immediate access to the network segment means limiting how many devices, computers and phones, can share a given segment.
So how many is too many? It depends. It depends on the speed of the network. It depends upon the number of simultaneous phone calls. It depends upon the amount of data traffic being generated by the computers. In other words, it requires regularly evaluating network bandwidth consumption by all the connected devices. Too much traffic combined with too little bandwidth means it is time to expand the network.
In the voice world, expansion of the overall capacity of the PBX occurs every time a phone is connected. Each phone is connected to the PBX via a dedicated line. No sharing. No waiting. No worries. In the data world, the capacity of the network segment remains constant regardless of the number of devices connected. Expand from 10 devices to 20 devices; the effective bandwidth available to each device has effectively dropped by 50%.
Thus we come to the issue of capacity planning. In the VoIP world, expanding capacity in order to accommodate additional devices requires adding network segments. Adding network segments costs money and personnel’s time. Whether or not these costs are less than those to expand the PBX must be evaluated on a case-by-case basis. There is also the issue of bandwidth usage for each device. Adding segments without regard for usage patterns may only produce a short term solution. Proper measurement of bandwidth consumption is necessary for expansion activities to produce a long-term solution.
Once the capacity issues that lead to the expansion issues have been worked out, there is one remaining hurdle to overcome; prioritization.
Line sharing in its simplest form treats all devices as equals. No one device has a priority higher than any other device when it comes to using the line. In the world of computers, that may be fine as delays are tolerated by design. In the voice world, delays are not acceptable.
When phones and computers share a line, it is important that phones be given access priority. The good news is that most vendors of data routers have developed some form of prioritization into their products. Interoperability of prioritization schemes between vendors is not a given. The only international standard for prioritization is included in a forthcoming IP standard called IPv6. Today, some vendors support IPv6 and others do not. The only way to guarantee end-to-end prioritization is to use routers from a single vendor or only use IPv6 for all the network devices. That may or may not be an option.
So does VoIP have a place in the enterprise? Absolutely. Remote agents are ideal for VoIP connections. It is usually cheaper to use a single connection for both voice and data than buying separate, dedicated lines for both voice and data connections. Branch offices are a similar situation as the cost of implementation is generally less than the cost of installing and maintaining both a PBX and data router.
Small workgroups within the enterprise are like a branch office in overall requirements. Segmenting their voice and data needs can be done with relative ease provided the term “small” is accurate.
Even large enterprises can be ideally suited for VoIP implementations. It all comes down to a thorough evaluation of traffic patterns, network capacity and cost.
The bottom line regarding VoIP is very simple. If you understand the underlying technology, have the tools to continually measure bandwidth consumption, have an end-to-end packet prioritization scheme in place and have network design expertise, VoIP can be a very viable way to deliver voice calls to the desktop. If you think a pilot project of 10 users translates directly into an enterprise configuration of 100 or 1000 users, let me be the first to say, “You have been warned.”