A totally non-actionable, useless, pie-in-the-sky thought.
Old fashioned desk phones are a bit like thin-client web terminals. A limited set of UI functions are communicated to a central, smart server. Management costs are minimised as a result.
The stupid network wants to invert all of this, and make the edges smart. When the telephony application is in a state of rapid flux, replacing a $100 device with a new, smarter $100 device is much less intimidating than risking the upgrade of a $10,000 PBX in the hope you’ve picked the right feature set. We’ve taken to using $1000 laptop PCs as telephones to get the functions we desire, because they’re more adaptable, and the change can be done with free or cheap software upgrades.
But there’s that little niggling issue of manageability of heterogenous distributed network devices. So I just wonder … could the idea of a “thin client” smartphone be feasible? Say you want a desk telephony device that can display Skype buddy presence, for example. Is the device itself the best place to put the Skype-specific application logic? Or is a device with a few softkeys, and a stripped-down web browser or Flash-type interface a better bet, interfacing to a server of some kind out there in the cloud?
There’s an unresolveable tension here. The stupid network is adaptable to change, and small incremental change at the edge is quicker than large change in the core; but change costs money, and co-ordinated change managed centrally can reap economies of scale. So as the level of uncertainty drops about what customers want out of Voice 2.0 over the next 5 or 10 years, expect to see the architecture shift accordingly.
Posted by Martin Geddes at 11:29 AMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/559.
Where application logic executes and where it originates are separable concepts. Scalability is usually acheived by moving CPU resources to the clients, however as Google is showing, it's in fact feasible to keep it in the server. Unfortunately, the knowledge of this black art is confined to only a few. That's why Google Talk's server based XMPP architecture should give everyone the willies.
Posted by: at September 14, 2005 03:34 PM...or in other words, there's no need to create a cheaply scaleable architecture if you're planning on making a lot of money!
The ultimate scarce resource is people, which is what makes manageability an issue that gently opposes the decentralisation trend.
Posted by: at September 14, 2005 04:13 PMintelligent Convenient Network Application Processors - scattered about, managed by the user, with morphed personalities driven by the specific user wants... its not about needs anymore.
Posted by: at September 15, 2005 04:41 PMMartin, remember the past 15 years of the mobile phone revolution. It began as centrally managed Intelligent Network. The IN portions have evolved rather slowly while an enormous effort (& $) have gone into ever-more sophisticated devices at the edge, i.e. mobile handsets. Yes, there's tension, but Voice 2.0 is likely to be more like Skype and less like Vonage.
Posted by: at September 16, 2005 07:07 AM