I'm afraid I must disagree with your stance on VoIP for the cable guys at this time.
Although this is an area that is probably better suited for the VoIP thread, maybe some of those guys can comment on the items below. I'll try to keep this concise so it doesn't take all afternoon to read. Much of this info is derived from articles on VoIP appearing in the August issue of CED magazine (thanks to them for the help).
In my mind, several key ingredients are still missing for commercial deployment of IP based cable telephony.
1. Improved voice quality
2. Redundancy
3. The full set of CLASS features like Caller ID (there are estimated to be close to 10,000 features on Class 5 switches of various permutations).
4. Reduced cable modem/multimedia terminal adapter (MTA) power consumption
5. Latency control
Given the above, the obstacles to broad-scale deployment of full-featured VoIP are fairly complex and will take some time to resolve. The most optimistic estimates I've seen still show another 18 months to go. The real world estimates typically double the "could be ready by" estimates making VoIP ready in 36 months. Only then do major trials start. Add another 12 months for trials and evaluations. We're now up to 4 years from today before VoIP is ready for deployment. Now add into the mix the following operational issues that may or may not be solved in that time frame to allow mass deployment:
1. How does the cable operator bill for an IP telephony call (circuit switched billing is well known, IP is not).
2. What is the optimum node size, given a 30% take rate for IP telephony (assumed take rate, no one really knows the demand for IP based telephony and guessing wrong on the node size creates lots of problems - just look at the cable modem problems as an example).
3. Using encoding/decoding compression schemes too many times -- as voice calls move from one IP network to another and/or are brought into voice mail) affects voice quality. Also, the G.711 audio codec, which uses a 64 kb/s bit rate could be a bandwidth hog (no one knows).
Let's for a moment take one of the primary obstacles to mass deployment of VoIP, that being latency (or round trip delay). This is a key factor in voice quality.
As an example, the Arris Cornerstone circuit switched system produces low delays of less than 5 milliseconds (ms) roundtrip between caller and receiver. Contrast this to VoIP delays right now of between 60 and 80 ms roundtrip. Rule of thumb states a voice call will break up at around 250 ms delay. Now in addition to the 60 to 80 ms delay must be added 50 ms of latency each time a router is encountered on a typical IP network. Since a typical US Internet user is 8 router hops from the core of the Internet, the latency additions would make voice conversations practically impossible. A private IP network of flat infrastructure with very few router hops is not only optimal, but essential.
Here's a big problem. While RoadRunner and Excite@Home's data networks on cable today may seem like natural conduits for cable operator's IP voice traffic,those data networks are not architected to provide seamless, end-to-end carrier grade quality of service (QoS).
Suffice it to say there are similar sticky wickets in all the other remaining issues shown above.
Now how about the numerous press releases that proclaim that VoIP is ready now, or just around the corner? Cisco is good at trumpeting rather loudly that VoIP works on a cable plant. The reality is that their recently completed technical trial with Charter Communications in Fitchburg, Wisconsin was unbelievably small. It involved just 5 (yes, FIVE) employees (not even general public) that Cisco issued a press release about announcing the "event". Um, I think I'll call the press when my neighbor's dog is paper trained.
Contrast this to AT&T, which has backed away from VoIP (with good reason), who will begin rolling out citywide circuit-switched cable telephony deployment trials over HFC in the following cities as we speak:
Portland, OR
Seattle, WA
Salt Lake City, UT
Denver, CO (all right! - some shameless cheerleading)
Dallas, TX
St. Louis, MO
Pittsburgh, PA
Chicago, IL
AT&T states the despite the advantages of IP telephony, such as routers costing much less than circuit switches, "we're not going to introduce IP until we think it's ready for prime-time". When, you ask?
According to AT&T spokesman Mark Siegel, not this year, not next year, but possibly in 2001, when AT&T will start deployment trials of VoIP.
If it's ready for trial by that time. Remember that 36 month interval at the beginning of this message? |
