About a decade or so ago, one theory popular among New Age telco pundits was the big dumb pipe. Networks really didn’t need all the fancy bells and whistles and complexity of IMS or quality of service QoS). Just make the pipes big enough and all network problems would be magically solved. It hasn’t worked out that way.
Part of the problem which remains today is that we can’t make pipes big enough at the core of the network to support high speeds from edge to core. The edge has become too fast.
Consider the initial movement over a decade ago to put 100 Mbps in every American home. It flopped and is now surpassed by a second wave effort to make gigabit speeds the next new standard for businesses and homes and everywhere. The problem with gigabit Ethernet speeds at the edge of the network is there’s not a lot of slack left when moving from edge to core. Ten Gbps speeds are the rough average of long distance links with 100 Gbps coming into play with select network investments.
And that’s it. There’s practically no upward growth path once you get 100 Gbps at this time other than adding multiple (expensive) 100 Gbps links. People talk about 400 Gbps and Terabit speeds, but that hardware is at least a couple years away for anyone but the deepest pocket research institutions and communications carriers.
Content delivery networks (CDNs) should have been the big clue that the big dumb pipe argument was flawed. Existing to effectively deliver popular content to the edge of the network without cluttering up core network links with tens of thousands of copies of the same stuff – yes, thank you Microsoft for the Tuesday update/bug patch cycle.
On-demand streaming media services – I’m looking at you Netflix and YouTube—have only exasperated the matter. Netflix works OK, but you can’t say the service works well all the time for all users. Delays occur and fingers are pointed between content provider and bandwidth last mile carrier, with the consumer stuck in the middle.
“Just make the pipes bigger,” growl the big dumb pipe people, “It’s nothing a terabit per second won’t fix.”
But we don’t have terabit pipes, or terabit speed servers or terabit to the home, and won’t have any of that hardware for commercial deployment in the next five years to a decade, give or take a couple of years.
Software defined networks (SDN) – adding intelligence to the network in an affordable fashion – are the only near term solution to deal with the many bumps and stumbles of today’s real world problems. And we’re still going to need a way to provide priority to real-time communications bits to make sure they arrive when needed in a competing world of video conferencing and streaming media.
There’s no simple answer to the ongoing challenges communications providers are facing in the decade to come as consumers are upgrading to gigabit links on the broadband side and tens to hundreds of megabits per second on the wireless side. The solution of the big dumb pipe certainly is one that doesn’t work in today’s network with any sort of near-term commercially available technology.
Edited by
Maurice Nagle
