The future used to be something you strained your eyes to see; now it's something you stumble over while racing from one innovation to the next. Innovations in information and communications technologies are happening almost as soon as we can imagine themFor a field with a long tradition of slow and steady but unexciting product development, there's probably been more hype about telephone systems in the last five years than in the preceding hundred. Voice over IP (Internet Protocol - VoIP), the death of the PABX, Internet telephony, so-called open systems . . . and that's not the end of it. Traditional and new developers of telephone systems offer an increasing diversity of products. There are more options for networking them. The future will be anything but dull. However, as always, prudence dictates that choices be made on business benefits, not supplier-generated hype.
Historically, evolution of telephone systems has largely been linear, with minimal architectural differentiation between the products of the leading vendors. There has been a relatively linear progression from electromechanical switches to stored program-controlled analog PABXs that arrived on the market in the late 1970s to the digital PABXs that first appeared in the early 1980s.
Today's PABXs are not radically different from those early digital PABXs. Sure, they are less expensive and offer more features, but these changes have continued to be evolutionary, not revolutionary.
The last few years, however, have witnessed the advent of two new categories of telephone systems, which, collectively, are significantly different from the established PABXs, key systems and centrex services. They are, therefore, a break from the historical pattern of evolutionary development. Although these new telephone system types will not replace conventional PABXs, they both present users with greater choice and they may prod the PABX manufacturers to increase the pace of product development from what, to date, can best be described as glacial.
The advent of PC-based telephone systems was almost inevitable. With the rise in the capability, power, cost-effectiveness and near ubiquitous use of PCs and local area networks, and a culture of dynamic and experimental PC product development that would climb a mountain just because it was there, the stage was set. At the same time, a contrasting culture of PABX product development at a glacial pace using almost totally purpose-built components gave further impetus. It was thus only a matter of time before someone would put these components together and voilà - the PC-based telephone system, also known as the UnPBX, was born.
Similarities and Differences
Architecturally, the PC-based telephone system is not that different from a PABX. It is typically constructed using a server chassis; telephony cards from suppliers such as Aculab, Brooktrout, Dialogic; analog handsets; a commercial operating system such as Windows, VXWorks or Unix; and the developer's own software.
So what is the difference? To begin with, the limits of the chassis and internal bus limit the number of circuits such systems can support, usually to fewer than 200 extensions. Because the companies developing them have typically come from a background of work with PCs and servers, product development is rapid and is not encumbered with the baggage of traditional ways of thinking. Such systems typically offer one or more integral capabilities and use a telephony window to provide the man-to-machine interface.
While in other respects the PC-based telephone system is a PABX, the telephony window is a paradigmatic change. Using a client/server application with a PC-resident client associated with each handset, the window displays information and provides functions for extension users making and receiving calls. Such functions are typically those provided by system integral handsets; however, without the constraints of the few functions keys and 20-odd characters of the handset's LCD, the telephony window has the potential to provide a far more comprehensive interface with the user.
Worldwide, there are more than 45 developers of such systems, only one of which, Interactive Intelligence, is represented in Australia (via Call Time Australia). Most of these suppliers have small installed bases.
Using many of the same components and coming from the same development culture, LAN-based telephone systems were also inevitable. As they use LANs to carry telephone calls, the LAN-based telephone systems are very different from both PABXs and PC-based telephone systems. Users make and receive calls with either a handset or headset that is connected to a telephony card in their PC or with a specific handset that is connected directly to the LAN. Gateways interface the LAN to the PSTN (Public Switched Telephone Network), ISDN (Integrated Services Digital Network), private network and/or peripherals such as a voice mail system or IVR (Interactive Voice Response) system. Such systems may or may not use the same LAN that is used for data.
Voice through the LAN
Although not all LAN-based systems use voice over IP - some use ATM (Asynchronous Transfer Mode) instead - most do and most of these comply with H.323, the international standard for the transmission of voice over such networks. H.323 defines the components, such as IP handsets and gateways, which collectively comprise a LAN-based telephone system. Compliance with this standard will allow components from different vendors to be used as part of the one system. On the other hand, the development of the H.323 standard is such that, at present, a multi-vendor system will have limited functionality compared to what is offered by single vendors with their proprietary extensions to the standard.
Architecturally, the LAN-based telephone system transmits voice between end-points in packets through a LAN instead of through dedicated circuits and timeslots. It is, therefore, a break from established telephony technologies. There are both advantages and disadvantages in this approach, yet it must not be assumed that because this approach is new and different that it is inherently superior.
As with PC-based telephone systems, LAN-based telephone systems also typically offer one or more integral capabilities and provide a telephony window.
Worldwide, there are about 20 developers of such systems, of which three are represented in Australia. Cisco distributors carry Cisco's CCN, 3Com sells their NBX-100 and Xter distributes the CosmoCom CosmoCall system.
Despite their advantages - the telephony window and integral capabilities being the most notable - PC-based and LAN-based telephone systems both have their inherent disadvantages, the first concerning their reliability. In contrast to PABXs and main exchanges which are reliable combinations of purpose-built hardware and embedded software, these new telephone systems largely use the standard PC and Windows operating system.
Although an improvement on a standard, garden-variety PC, industrial-grade PCs, which at best will comprise only part of the overall telephone system, fail to match the reliability of a PABX. Then there's the Windows operating system. Designed to be used from applications ranging from computer games to industrial process control to office desktops, Windows is probably the most complex piece of software ever written. However, software complexity and reliability are inversely proportional. Hence, a telephone system using the Windows operating system cannot come close to the reliability of the software which is designed to perform the single function of operating a PABX.
Loss of Mains
The other shortcoming of both PC-based and LAN-based telephone systems is their susceptibility to a loss of mains power. The whole culture of the PC/server/LAN environment is one of using uninterruptible power supplies (UPSs) with capacities in the order of 10 to 15 minutes, just long enough to do an orderly shutdown. The telephony world, by contrast, has a tradition of having switches powered from 48-volt DC batteries with capacities ranging from four hours for most PABXs to several days for telephone exchanges. Although there is nothing preventing developers of PC-based and LAN-based telephone systems from having them powered from 48-volt batteries, collectively they are as reluctant to introduce this development as PABX vendors are to develop a telephony window.
Although PC and LAN-based telephone systems have come with some hype, nothing matches that generated by the claims for voice over IP. Voice over IP is not only being trumpeted as the next big thing, much of the promotional material strongly implies that an organisation that fails to adopt this technology fast enough will risk missing the boat and face imminent bankruptcy. What nonsense.
There can be no doubt that IP is almost ubiquitously used in corporate LANs and WANs, and the techniques and products developed for it to carry voice are now established and widely available. Nevertheless, just because IP networks can carry voice, to assume that they soon will be carrying just about all of it is a ridiculous exaggeration.
Contrary to what is implied in material promoting this technology, voice over IP is not a singular technology. There are actually four applications of VoIP, the economics of which are quite different. They are:
LAN-based telephone systems, as described above;Carriage of voice by a private IP network;Carriage of voice through an Internet-based virtual private network (VPN) between points of presence operated by a national or international ISP; andCarriage of voice between two PCs, typically operated by individuals, through the Internet.
Compared to the other transmission technologies of ATM and frame relay, IP is ubiquitous and its components are thus of modest cost. However, there is no priority bit or byte in the IP packet header; hence, voice can be prioritised only at the point of ingress to the IP network. Although most voice-capable routers can do this, as an IP packet carrying voice traverses a WAN, it will be just as susceptible to delay as any other packet. With careful engineering, IP networks can carry voice and maintain quality of service; but to do so, the network must have been adequately dimensioned, and the traffic volumes must be continuously monitored by staff who understand both LANs and voice.
There are also the problems introduced by interfacing PABXs to packet switching networks - whether IP, ATM or frame relay - the most significant of which is congestion. Consider the situation where there are spare circuits between the originating PABX and its local gateway but none between the remote gateway and its PABX. The call will be connected to the local gateway and forwarded through the network to the remote gateway, but will not be able to be connected to the remote PABX. As this congestion will occur only after the first gateway has received the call and routed it through the network, it will be too late to busy the originating PABX's circuits which would have allowed the call to be alternatively routed by that PABX through the PSTN. The caller will thus receive an engaged tone when the remote extension may not be engaged.
Another problem concerns the transmission of QSIG, the international standard protocol for interPABX signalling. QSIG allows transparency of features such as callback and display of calling extension number between dissimilar PABXs. QSIG and other such protocols are designed to work on point-to-point links by which a call exiting one PABX's channel 12, for example, is received on a specific other PABX on channel 12. With a packet-switched network, a call exiting one PABX's channel 12 could arrive on any other PABX, on any channel. For QSIG to be able to provide extension feature transparency, some sort of protocol transparency unit is required.
The 2000 edition of Telsyte's Industry Profile: Australian Data Service Providers reported that interviews with 28 managed IP providers indicate that eight (29 per cent) offer a managed VoIP service and another seven are planning one. However, in her article "Voice over IP - Hype or Reality" (CommsWorld, April 2000), Telsyte managing director Shara Evans found that, for all the interest, there were few real-world production applications.
Before any organisation assesses the merits of voice over IP in comparison to voice over any other technology, the question must first be posed of whether a private network is really the best option. With the cost of long-distance calls having declined significantly over the past decade and the advent of virtual private networks - carrier arrangements which combine substantial call discounts and some features such as an intra-organisational dialling plan - a private network may not be the most cost-effective means of carrying long-distance calls.
Telsyte's Industry Profile: Australian Data Service Providers also reported an estimate by 18 ATM service providers that 18.6 per cent of ATM traffic consists of voice transport, and an estimate by 17 frame relay service providers that 28.4 per cent of frame relay traffic consists of voice. Although, no doubt, growing, clearly, a lot of voice is still carried by TDM (Time-Division Multiplexing) networks, VPNs (Virtual Private Networks) and the good old PSTN.
Another factor driving the development of PC and LAN-based telephone systems was the historically primitive and arcane facilities which have traditionally been used to configure and administer PABXs - facilities which have been neither user-friendly nor intuitive.
Although PABXs now typically offer more user-friendly, Windows-based software for switch configuration, vendors of PC and LAN-based telephone systems have argued that their systems can be administered by an organisation's existing IT staff with PC and LAN expertise. This is, however, a risky proposition.
While such staff may not have to understand the complexities of administering a PABX, they must still understand telephony. Nothing illustrates this more than the potential consequences of mis-administration. A PABX or PABX network that is not competently administered may have too many trunks or tie-lines, which will waste money; or too few, resulting in occasional congestion and blocked calls. But quality will be maintained. However, a LAN-based telephone system, or indeed any telephone system using voice over IP, if not managed competently, also risks congestion. The difference is that with such a system, congestion will have an impact on call quality.
Still on administration, another area where PABX suppliers have made changes in recent years concerns digital extensions and extension management. The prices of digital extensions and handsets have decreased from their introduction to a point where the use of only digital handsets imposes a small cost penalty. The use of only digital extensions allows moves and changes to be administered only by reprogramming from the PABX's administration facility without expensive cable re-jumpering, ensuring cable records remain accurate.
However, there's still work to do. When a staff member joins or leaves an organisation, they are assigned a PABX extension and their name is added to the PABX configuration. A parallel change must be made to the telephone information and management system (TIMS). Unfortunately, there is no industry standard interface that would allow a single update function to change the configurations of both a PABX and a TIMS.
So what does the future hold? Although forecasts in this industry are often well off the mark, it is reasonably safe to assume that PABXs will still be with us for years to come. They will continue to be developed, albeit slowly, and developments will continue to be driven more by what the manufacturers want to develop than what the market really wants. Nevertheless, PABXs are certain to continue to be the highly reliable systems to which we are accustomed. If presented a choice between dynamic feature development and reliability, the market will continue to overwhelmingly choose the latter.
Stephen Coates is a Sydney-based independent telecommunications consultant. He can be reached at email@example.com
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.