Chris Kranky

Recent Posts

Cloud computing for telecom applications?

Chris KoehnckeChris Koehncke

Amazon Cloud computing for telecom? Is that a real possibility? I’m
not talking about a “Verizon” offering cloud computing, but rather be a customer
of Amazon ECS. Logically it make sense, but will they see the light? Or will
someone else have to show them the way.

As voice communications moves toward IP or VoIP the engines
that manage the “switching” of these telephone calls are nothing more than
software running on general purpose computers with Linux as the core operating
system. The way VoIP operates is all the packets for those calls don’t have to
run thru these “switches” the voice portion goes from gateway to gateway. The
only IP packets these VoIP “switches” see are those associated with signaling,
more akin to the SS7 network of olden days.

Cloud computing probably isn’t the best selection for an
application that runs 24 x 7 x 365, while the economics from Amazon (and
others) are starting to make this attractive on a pure dollars basis, but there’s
a bigger savings here to be found.

Verizon Wireless has ~ 90 million wireless subscribers
throughout the United States spanning 3 time zones. Verizon has at any hour of the day, billions of dollars
of servers and equipment ready for those 90 million subscribers to start making
a telephone call. But at 2 a.m. in the morning, only a tiny fraction of
these subscribers have their phone on, much less engaged in a phone call. Yet all of Verizon Wireless switches are on, consuming energy, emitting heat and basically doing nothing.

The first question is whether Verizon Wireless could
successfully architect the capability to shift a subscribers serving server on
a demand basis. Thus at 4 a.m. Pacific Coast time, the sleepy California servers could pitch in to help the East Coast during the morning commute and vice versa as the sun moves
to the west.

But take this a step further, why does Verizon Wireless need
computing capacity to switch calls for the peak period of say 1 or 2 hour time
periods during the day. Couldn’t they utilize cloud computing to “fire up”
additional server horsepower on a demand basis. Similarly, if Christmas falls
on a Monday, they could not need the additional horsepower on that day and not
fire it up. Depending on how fast these additional resources can be brought on
line, they could even elect to fire up the cloud only when they foresaw the demand.

You could quickly imagine the significant reduction in
capital expense right out of the gate, plus factor in savings on power and data
center space and then whack a few telephone company employees for a twist of
fate. This sounds pretty good.

Telecom architectures and flexibility have never been words
used in a single sentence and VoIP, for the most part, has continued to build monolithic
architectures on space age networks, but the potential exists for some to focus
on cost savings as the primary motivator of VoIP and not some new feature.