TDMoIP

There’s a new technology to learn about… TDMoIP.

VoIP (Voice over IP) has been a big disaster. Lots of engineering effort in, little success out. I had a Vonage phone line for a while, and the call quality (mostly due to latency) was terrible. I was using my Vonage endpoint under significantly less than ideal circumstances, so it might not have even been their fault. But I’ve seen VoIP stuff take up lots of time and deliver very little value at work too.

The problem for VoIP is that it is competing against the rock solid, “can hear a pin drop”, telephone network. The way the telephone network makes such reliable, low latency, quiet phone calls is that it uses a very inflexible but deterministic system called TDM, or Time Division Multiplexing. TDM is expensive, because it can be wasteful. VoIP strips huge amounts of seemingly antiquated TDM technology away, which seems like a breath of fresh air to Internet-oriented engineers who are asked to implement VoIP. But then it turns out it is really hard to implement reliable telephone service, and throwing away all the work Bell Labs did to make TDM work so well wasn’t such a bright idea.

So, TDMoIP (which really needs a sexier name… “packet telco” maybe?) is basically an attempt to use existing, trusted techniques to pound the square (reliable) peg of TDM into the round (cheap) hole of packet networks.

By the way, another good example of reliable and deterministic technology that beats complex technology is orbit delivery systems. The cheapest way to get to orbit is still on expendable liquid and/or solid rockets, not via the Space Shuttle. And rockets are more reliable too, on a chassis-per-chassis basis. We’ve built five shuttles and lost two. We’ve built thousands and thousands of rockets, and lost significantly less than 40% of them (before their designed-in destruction, of course). Yes, I know this is not a fair comparison, but it’s an interesting datapoint to start a conversation with, wouldn’t you say?