I've been at a couple of conferences this week - covering Telco 2.0 business models and femtocells. At each one, an "old school" telecom speaker has made reference to "Telco Grade" infrastructure and capabilities, either in terms of security models or reliable voice services.
This isn't new - I always hear representatives of incumbent operators or traditional infrastructure vendors talking about "Five 9's" and QoS. Voice quality scores (MOS) get quite a few mentions too.
It made me start wondering whether these concepts, which are heavily rooted in a world where telecoms was all about voice sessions and dialtone, are in need of a serious update.
In particular, any future metrics for reliability and quality need to cast their net much wider. Marek Pawlowski over at the Mobile User Experience has a coruscating post about his nightmare experiences in buying a T-Mobile Android G1 phone, for example. The fact that ordinary circuit calls might work quite well is clearly almost irrelevant to his view of either the "quality" of the operator or the phone itself.
Similarly, if we're talking about IP services like VoIP, IM or even full IMS over a mobile connection, the fact that one link of a long chain has managed QoS (and even differentiated billing) is totally irrelevant if 30% of the time I'm outside decent 3G coverage. That's the gating factor on quality, not what's going on in the middle of the NGN - you might as well just use the Internet.
It's like posting a letter from London to New York, and getting a QoS guarantee that the central London to Heathrow Airport leg of its journey is 99.999% certain to take 47.2 minutes. But there's a 30% chance the flight might get delayed until tomorrow, and another 17% chance the letter will get sent to Boston by mistake. And a 75% chance the guy at the Post Office sold you the wrong stamp in the first place.
It's also critical to think about metrics other than basic coverage and availability when it comes to "mobile Internet grade". Latency, jitter, packet loss and speed of connection setup can be critical for many applications. And it's amazing that many of the so-called "telco grade" mobile broadband networks have distinctly non-Internet grade DNS lookup capabilities. Loading a MySpace page with all its plug-ins might mean resolving 70+ IP addresses, and again, if you do it over mobile the critical factor isn't the operator's core network.
The bottom line is that vendors and operators continuing to use the "carrier grade" and QoS arguments are often missing the point, or are being disingenuous. Why over-invest in "quality" in one part of the network, when basic radio coverage, IP network internals, or customer service are the "weakest links" on your customers' overall perception of service reliability and performance?