Voip or h.323 or ip telephony
Voice quantity or voice quality? With 75 per cent of backbone voice traffic likely to be IP within five years, are we sacrificing voice quality for traffic
VOICE QUALITY HAS been the Achilles' heel of putting voice traffic over a data network. While no one will balk at an email arriving a few seconds after being sent by a colleague on the same LAN, a telephone call with delay, echo, drop-outs and `dalek-like' overtones will not be tolerated.
Early VoIP implementations, particularly those designed to shunt international calls over the public internet for hard-up world travellers, suffered outrageous voice quality issues. Since this time, a great deal of work has been done on improving delay and on traffic prioritisation in dedicated networks--if not the internet. While this has significantly raised the quality standards of VoIP calls, it is still to improve a perception built on a back-packer's budget.
"Voice quality will always be open to interpretation by the recipient," says Ian Boreham, marketing manager at ONI, "although today's VoIP and IP telephony solutions have an extremely high quality of voice. Whilst this may not be true `Toll' quality it is very hard for the average user to tell the difference."
Certainly perception of voice quality is subjective and what sounds good to one person may sound awful to another.
Any major delay, jitter or latency is unacceptable in a business orientated IP telephony environment, and it is generally accepted that any delay over about 120ms on the LAN is audible to the human ear. Thankfully, within a controlled LAN/WAN environment at least, current technology can keep voice traffic moving a lot faster than that.
Equant's VoIP specialist, Jon Floyd explains: "Class of Service (CoS) in the access layer and Traffic Engineering in the core, using MPLS in conjunction with IP Diffserv, has overcome many of the initial problems of VoIP. When supplemented by packet interleaving, large File Transfer Protocols (FTPs) no longer have the negative effect they did on high priority, time sensitive applications such as voice. If these are backed up by voice specific SLAs from the operator, corporate users can be assured of high quality business grade VoIP," adds Floyd.
Malcolm Needham, pre-sales manager at Telindus, argues that voice quality is dependent upon a number of factors, not all of which are network related: "The quality of the handset used or, in the case of an IP Softphone, the speaker/earpiece and microphone within the PC, all play a large part in overall voice quality.
"Moreover, utilising VoIP applications across a WAN can be complex to setup and manage, depending upon the types of WAN technologies employed and the QoS facilities available on them," adds Needham. "In an ideal scenario, with limitless bandwidth and therefore negligible congestion, VoIP applications would require no special handling on the network. However, in real world environments, QoS mechanisms are essential when deploying either IP telephony or Toll bypass type VoIP applications."
IP TELEPHONY TAKE OFF?
There have been many false dawns for IP telephony, not least of which born of the dot.com fuelled chaos of the late 90's. "A problem we have as an industry is that VoIP was amazingly hyped some three years ago," says Andy Trott, manager for Pace Vegastream. "Everyone thought it was going to be the new telecoms standard and people invested heavily into it ... and lost a lot of money.
Perhaps this renewed interest in VoIP reflects a market choosing voice and data systems with far greater consideration than ever before. "The `what will we do with our phone and data network' issue is moving into the realm of senior business decision-makers," says Richard Pennington, senior consultant at Lynx Technology. "As a result, the breathless `won't-the-future-be-wonderful' techno-evangelism is making way for level-headed questions about current kit's capability and sober analysis of how others have done it. Last year, tryers were looking in the crystal ball; this year buyers are looking at their bottom line."
But perhaps the biggest reason why IP telephony is on the up is good old fashioned proof of concept. The fact that the technology is `here and now' means that user names can start speaking louder than vendor claims.
Toni Fabbricini, marketing manager for Westell, takes this pragmatic view one step further citing that keeping an eye on the big players is always a good indication of the direction of the herd: "One thing to remember is that the likes of Nortel, Lucent, Avaya, Siemens and Ericsson have all made the commitment to VoIP. Therefore, logic says that if nobody is developing new TDM technology then packet will win because in the not-too-distance future there won't be much else to buy. We've come a long away in a very short space of time and I don't believe there is a way back."
So can IP telephony offer PSTN quality with added functionality at a fraction of the cost of a traditional voice network? "Yes it can in terms of voice quality, but not yet in terms of ubiquity," says Simon Boyle, consultant at Dimension Data.
"VoIP is still a business tool for enterprise customers. True, some of the more imaginative ISPs/carriers are developing hosted VoIP offerings (the centrex model) and this is interesting. The business case for having just one IP based pipe that delivers all your voice and data services is very attractive," adds Boyle.
Anurag Patel, managing director of the Techland group, similarly sees the benefits of VoIP against the PSTN--but not necessarily on the strength of call cost savings: "We do not consider VoIP as a `one size fits all' solution. For example, we have found that some very large corporate customers have negotiated extremely competitive flat rates for voice calls with their incumbent carrier provider. This means that the additional equipment expense and time required to deploy a VoIP infrastructure mean that it is not cost effective."
"There are good and bad aspects associated with both VoIP and the PSTN," adds Grant Notman, UK General Manager at RAD Data Communications, "but I think it is fair to say that PSTN still has the edge. Everyone can demonstrate a perfectly functioning system with their own kit. It is when they have to interoperate with anyone else's that the issues arise. In the majority of cases, at best you are looking at a VoIP hybrid solution which is going to affect the quality due to the conversion involved."
VOICE OVER BROADBAND
For corporates selling to consumers (b2c) and consumers themselves, the concept of a low-cost telephony system with all the added features of an IP network is very appealing. In the US many vendors are looking into Voice over DSL and multiplexing out several virtual phone lines from a single copper pair--ideal for small businesses or big families.
But current broadband connectivity options, such as ADSL and cable, offer no facilities for QoS. Whilst many users have found they can implement VoIP across these networks today, as their usage increases this is likely to present more of a challenge.
"ADSL is a contention based technology offering a best efforts service with no quality of service or traffic prioritisation concepts," says Di Data's Boyle. "Our experience to date is generally that there is sufficient speed to overcome these issues and VoIP will run over ADSL. However, once the traffic levels pick up most of the currently available contention ratios (50:1 or 20:1 for business users) will make VoIP over ADSL a bit hit and miss. The other issue with this approach concerns security and the need for secure tunnels through the internet for voice traffic. IP Virtual Private Network (VPN) technology meets this need and lab tests show this approach has some merit."
Although VoIP is still a way off being a truly all encompassing end user proposition, there is no denying its tremendous impact in the backbone. A recent Analysis report estimated that IP is already being used to carry about six per cent of all international traffic on routes where competition is limited. The study also found that that IP telephony was becoming a serious consideration for corporate communications managers because of the expected savings in capital costs, per-call costs and system management costs.
But the road to VoIP is not a clear cut path according to Analysis' IP specialist Margaret Hopkins: "At the moment, VoIP's most compelling advantage is the cost-effectiveness with which small sites can be connected to a corporate voice network. However, there are now three competing protocols for end-user VoIP terminals--H.323, MGCP (media gateway control protocol) and SIP (session initiation protocol)--and this will slow down market development."
While VoIP is still not ubiquitous, problem free or simple to implement, the issue of voice quality in a managed network is dead and buried. In the words of Hopkins: "Voice over IP is finally becoming capable of substituting for circuit-switched systems, and on managed networks can match or even better voice quality."
CASE STUDY
VoIP in a bottle