Voice Solutions

Remember earlier in this chapter when we mentioned that security is an extremely large topic? Well, the convergence of modern data and voice networks is not far behind in scale. Cisco voice certifications? Check. Entire books on IP telephony? Check. Entire careers in IP telephony? Check. Once again, we’ll be scratching the surface of a very deep topic. Once again, we’ll be discussing this technology in light of how it impacts modern network design.

Let’s begin by discussing some of the traditional ways voice traffic has been handled. Then we’ll look at the implications on the design process of integrating voice and data networks.

Traditional Voice Technologies

As humans, our voice communications are analog. Our vocal cords and ears are preconfigured to produce and receive analog voice communications and work quite well. However, there are distance limitations in our natural equipment, so the telephone has emerged over the last hundred or so years to help us communicate at greater distances.

Analog communications are not as desirable on telephone networks as they are to human ears. Analog communications are subject to distortion when amplified and are not efficient users of transmission media. Converting analog conversations to digital (and back again) allows digital signals to cross the telephone network, and this overcomes the previously listed issues with analog communications.

Analog signals are generally converted into PCM (pulse code modulation) digital format signals. This analog-to-digital conversion process requires several steps, including filtering (eliminating sounds outside standard frequencies) and sampling (at a consistent interval).

PBXs and PSTN Switches

There are two types of telephone switches. They have many features in common such as the ability to connect multiple telephones and route calls between them. However, there are some differences as well.

Private branch exchanges (PBXs) are essentially telephone switches used by private enterprises, where public telephone switches are owned and operated by telephone companies. PBXs are used to implement many features within an enterprise such as voice mail, call hold and music on hold, call transfer, call parking, conference calling, and call history. Within an enterprise, multiple PBXs may be interconnected, and internal calling and telephone services will work without any connection to the public switched telephone network (PSTN) at all!

PSTN switches are more, well, public than PBXs, and fill a different role. First of all, where PBXs generally support thousands of phones, a PSTN switch supports hundreds of thousands of phones. PSTN switches are used to connect individual telephones, PBXs, and other PSTN switches.

PSTN switches are generally located at a central office (CO). For example, you most likely have a telephone in your home. That phone is connected to a PSTN switch in a CO via a local loop. The local loop is the physical cable between your home and the CO and is used to provide your phone line. There are connections between other types of devices. Tie trunks connect PBXs, CO trunks connect a PSTN switch to a PBX, and PSTN switch trunks connect PSTN switches.

All of these devices must have ways of talking to each other; that’s called signaling. In the broadest sense, there are two types of signaling: that between a switch and subscriber (subscriber signaling) and that between multiple switches (trunk signaling).

PSTN Services

Modern PSTNs offers a variety of services to home and business users. We mention a few here that are relevant in converged data and voice networks:

Voice mail Ah, who doesn’t have a love/hate relationship with voice mail? As a service, it truly is an indispensable part of business voice systems today.

Centrex Centrex services are essentially having the PSTN act as if it is your private PBX. There are numerous advantages to this over actually purchasing and installing a PBX. Centrex is an outsourced service, so there is no infrastructure to buy, house, and maintain. Service is provided based on monthly fees. Services such as call forwarding and transfer, three-way calling, and closed dial plans are available.

Call center The call center concept is based on the automatic call distribution (ACD) system. Most people have called a call center and experienced the customer perspective of ACD. Incoming calls are accepted; if no agents are available, the incoming calls are “buffered” and music is played while you wait for an agent.

Interactive voice response (IVR) Did you cringe? If not, you have never called into an interactive voice response (IVR). IVR systems are responsible for that endless cycle of “Select 1 for support, select 2 for sales, ….” A very efficient system, it allows information to be retrieved via telephone lines without anyone on the other end of the line. Certainly a vital business feature, but in the author’s humble opinion, a bit overused by some.

That is a quick introduction to the infrastructure and services available on traditional telephone networks. It is only recently that people have been converging voice networks and data networks effectively, and there are many choices available. Most new data networks need to be evaluated for the possibility of supporting voice traffic, if not immediately then eventually.

Now let’s look at some of the issues that arise when we migrate this mature voice technology to the rough-and-tumble world of data networking.

Integrated Voice and Data Networks

So what is the big deal with converged networks? There is no single answer to that question, but if there were, it would likely be cost. Consider a medium- sized enterprise with multiple sites. It is not uncommon in such an example to have multiple PBXs, connected by leased tie-lines. It is also not uncommon to have multiple routers connected by WAN circuits. Two separate networks. Two bills. Two teams of engineers to support them. This is one area where convergence really shines. The ability to get a single network leads to cost savings in the long run. There is really nothing wrong with PSTN, but if data, voice, and video are going to converge onto a single network, it can’t be the PSTN—it must be the data network.

Security and application delivery are other issues in converging voice and data. There are many options for security available on IP networks, and it is easier to secure a single network than multiple networks. Application delivery to voice solutions on data networks can move quicker than on traditional voice networks.

Certainly the most popular technology to integrate voice and data onto the same network is voice over IP (VoIP). As the name implies, voice traffic is digitized, placed in IP packets for transport across IP internetworks, and then converted back to sound when it reaches its destination.

However, VoIP traffic crossing a Frame Relay link is not the same as Voice over Frame Relay (VoFR). With VoFR, a voice-enabled router might be directly attached to a PBX. The router takes a voice feed from the PBX and converts it to Frame Relay frames (no IP) for transport. The same applies to Voice over ATM (VoATM).

H.323

H.323 is an ITU-T defined protocol capable of carrying audio, video, and even data across IP networks. H.323-compliant devices are theoretically capable of interoperation. H.323 established standards for compression and decompression (codec) of voice (and video) traffic, allowing devices from separate manufacturers to interoperate. Since it runs on IP, H.323 is capable of traversing the Internet as well as private IP networks without regard to implementation details.

IP Telephony

While H.323 is a specific protocol used to carry voice traffic across IP networks, IP telephony is the structure and service that facilitates this communication. Essentially, the IP telephony architecture defines how to remove a PBX and replace the services and functionality using IP networks. IP telephony includes four components:

Infrastructure The infrastructure is the component required to interconnect all devices. Phones (endpoints) are connected through IP-enabled routers and Layer 2 switches to the broader IP network, as well as with the PSTN network.

Call processing The call processor is the “brain” or central component. Operating much as the PBX does, it handles, well, call processing. Cisco CallManager (CCM) is Cisco’s call processing offering.

Applications Applications are the functions similar to the PSTN services discussed earlier. No one wants to lose functionality, and migrating to a converged voice and data network requires the replacement of existing services on IP. IVR, call center (ICD), voice mail, and automated attendants are all available applications.

Client devices These are the phones. Cisco has both hard phones (real handsets) and soft phone offerings. Soft phones run on a PC as software.

Voice Issues

Let’s face it, data networks were not initially designed with voice transport in mind. While they do work quite well, managing data networks with voice traffic requires attention to a few details that typically fell “below the radar” when managing data traffic. Let’s discuss several of these issues.

Delay

Delay is a major stumbling block for voice traffic and needs to be addressed with the various QoS mechanisms discussed later. Distance is usually the major contributor to delay in existing voice networks. In a phone call to a friend across town, the delay due to distance is imperceptible as the electrical signals travel at the speed of light. In a phone call to someone 8,000 miles away, however, the delay can be noticeable. Propagation delay is the time required for the signal carrying voice traffic to travel the distance across the physical network medium. When distances are short, propagation delay is negligible. As distances increase, delay increases also.

In integrated networks, delay can be a voice-quality problem. Voice information has a characteristic timing. A user will utter a particular syllable of a word with an interval of time between it and the following syllable. Since this tiny pause is as much a part of speech as the verbalized parts, preserving its timing is essential. In traditional voice networks, the voice channel is a synchronized bit stream that preserves the timing of all speech elements precisely. However, in data networks, inserting delay due to congestion or handling corrupts the speech.

Constant delay should not exceed 150ms in one direction. Anything above 400ms renders the network unusable for voice traffic. Constant delays can be caused by processing delay, serialization delay, or propagation delay. However, not all delays are constant. Next, let’s look at variable delay.

Jitter

Jitter is inconsistent delay, and it can be very annoying. It is caused by a number of factors, including high network utilization and queuing problems. Jitter can be compensated for to some degree through the buffering of packets. Since packets are received at an irregular rate, they are buffered and played back at a constant rate. The dejitter buffer handles this task.

Packet Loss

A voice conversation on a network is a stream of packets. Each packet represents a small time slice of the voice conversation (20ms). Each lost packet represents a “skip” or “blank” in the conversation. While codecs (compressor/decompressor) can generally deal with the loss of a single packet, the loss of multiple packets can cause audible gaps. With data traffic an upper layer protocol would simply notice the lost packet and retransmit, but this is simply not an option when the payload is voice.

QoS

QoS stands for Quality of Service. While broad in its implementation, the basic concept is that of marking voice traffic as “important,” and then letting “important” traffic have a right-of-way through network devices. When implemented, QoS can allow voice traffic to speed through normal data network congestion. In order to work, the voice packets must be marked or “colored” to indicate that their content is time-sensitive. Intermediate network devices must be configured to look for this mark and then queue the packet appropriately. QoS is an effective tool in dealing with delay.

That introduction brings us to the end of our voice discussion. As mentioned previously, and like security, this is a very deep field and we haven’t begun to explore it here. However, we’ve covered the concepts of legacy voice and its migration onto the data network. For more details, check out CCO at www.cisco.com.