TDM and VOIP

May 22, 2006

By David Greenfield Networking Pipeline

There's plenty to like about VoIP. There's also plenty to hate. While VoIP vendors are good and touting the former they tend to ignore the latter. Here then are six "gotchas" picked up over the years of deploying and reporting about all things related to VoIP. None of which will eliminate the technology from your business' consideration, but all of which must be accounted for when you consider the technology.

1. Reliability Headaches - It's a fundamental axiom of manufacturing engineering and just plain common sense that all things being equal, the product with fewer parts will be the more reliable one. If that's true then VoIP starts out with a significant strike against. Unlike digital networks that just telephony switching equipment, cabling, and telephones, VoIP systems brings the additional complexity of the underlying data network. The choices made at those layers, independent of the VoIP network, will make or break a voice system. Mess up on the VLAN or QoS settings of your switch or router and watch voice quality tank the next time a virus-infected PC spews junk onto the network. Reboot a router or switch and watch voice calls drop. Lose power and in properly designed data network will send voice to you-know-where.

2. Hidden Costs – A two- or three- year old 100 Mbits/s or even 10 Mbits/s infrastructure may have a fair amount of gas left in the tank, but the conversion to VoIP will force most companies to buy new data equipment and even new wiring. They'll need switches that implement 802.1p (MAC-layer QoS) and 802.1q (VLANs). If availability during power outages is necessary then those switches will likely need to be equipped with inline power (802.3af). All totaled expect new edge switches to at least run $68 for a 48 port switch. A VoIP-enabled router, like HP's ProCurve Secure Router 7000dl, will run $1,173 to $2,303 depending on configuration. This says nothing for fees relating to network design, installation, and maintenance contracts

3. Plenty of Insecurity – VoIP bigots will remind you that wiretapping in the PSTN is a technically trivial task for any accomplished technician. What they won't mention, however, is that while the PSTN carries it's own vulnerabilities, the VoIP world must contend with a potent combination of global accessibility and simple-to-use hacking tools. Users otherwise ignorant of the ways of wiretapping or war dialing can easily download applications, such as Cain and Abel or Trinity and snoop on phone calls or launch a massive distributed denial of service (DDOS) attack that disable a phone number. Companies looking at deploying VoIP must plug those holes in their data network if the voice network is to be secure.

4. Talking Underwater – Digital telephony is nice and neat. Pick up the telephone, dial, and talk. Voices never sound as if they're coming from the handsets filled with seawater. Yet run VoIP over a poorly engineered network and you'll find warbled voices, echo, and just about any other type of distorted voice condition.

5. The Saga of the SoftPhone – Some aggressive VoIP providers will argue that the PC can be used as a telephony replacement. Just buy a USB headset, load up a soft-phone, and walla! Companies can save the cost of their digital handsets. The fact is though that most companies will find their soft-phone performance to be erratic. The combination of Windows XP and PCs have yet to provide an effective way to prioritize voice threads, allowing applications to hog the a PC's processor and wreck havoc with voice calls. Echo cancellation is difficult to tweak when the two soft-phones on either end of a call come from different manufacturers or when a softphone communicates through a PSTN gateway to a regular phone. Much of that might change when Vista ships and with improvements making in processor design

6. Brainteaser – What size access line does it take to service 114 telephones on a VoIP network? Don't fret for not knowing. Voice engineers spend their lives noodling around with those sorts of questions. They're difficult enough in the digital world, but in the VoIP world they become even more complex. Is this a trunk to a VoIP network the PSTN? If it's to the VoIP network what type of CODEC do the telephones run? What's the underlying transport network? What other applications are sharing that access line? Voice engineering across a WAN of any sort is a rather complex science one that’s only made more complex when the network is VoIP.

In the end, most companies move to VoIP for reasons of attrition. They've written off their PBXs and its time to for an upgrade. Often they're find that the costs of paying for the existing maintenance contracts can come close to matching the cost of a new IP PBX. There are plenty of other good reasons for moving to VoIP just few that are compelling enough for most offices to make the switch on their own merit.