Treading the exhibition floor, here’s my gut feel on the buzzword trends at VON based on having come for the last few years.
I had an interesting conversation with Brix, who make embedded measurement services for everything from core networks to end user equipment. They’re in competition with a whole bunch of other vendors here like Tektronix and Ixia. I’m kind of fond of these kind of services. They don’t break the end-to-end model of the Internet (“read-only intelligence”), and you can see them prosper regardless of whether the network is best-effort or QoS based. A real Telco 2.0 business. You can imagine an enterprise in a couple of years deciding whom to buy WiFi roaming from based on the typical quality of experience of real users on those different networks. Brix win regardless of the direction the telecom industry takes in terms of open or closed networks.
I also had a chat with Zyxel, who make a variety of end user equipment. Among other things they make a WiFi phone, which has a rather stiff price but is mostly targeted at verticals like schools, warehouses and hospitals. It’s one place where I personally haven’t looked very hard for “Voice 2.0” features, but you suspect that there are vertical customisations to be made. Will we end up with many islands and archipelagos of VoIP, each with subtly different functions, and if so how will they interconnect?
At the moment I’m sat in the session on IMS Lessons Learned. The lesson is clear: it’s chaos! Everyone is implementing the edge devices with subtly different semantics and protocol variations. Different network equipment vendors do things differently. Every part of the puzzle has multiple standards to choose from. It makes Unix’s Tower of Babel look like a monoculture. And we all know how that ended: a private, unified standard — Windows — won the game.
One example given is overload events, such as a regional power outage and resumption making all SIP phones re-register again at once, launching an unintended denial of service attack on the host voice network. Skype insiders will be chortling at this point: Skype clients stay authenticated over temporary disconnection, and they communicate presence data peer-to-peer without being throttled by a central network function.
Verizon’s pitch is a bit more sensible: only implement stuff that generates cash quickly, only implement the bits you really need to do that, and don’t spend too much money selecting lifelong vendor partners as you’re probably going to sleep around a bit. Caveat investor.
Posted by Martin Geddes at 07:31 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/792.
I agree. Even deregistering and reregistering when migrating from 1 access network to another may not be acceptable for the customer.
I also wonder if any benchmarking test results are available for the CSCFs when subjected to a million subscribers' SIP signalling load.