Mix and match

Editorial Director and Associate Publisher Stephen Hardy points out that not only are no two networks the same, but frequently no two parts of the same network are identical. Therefore, the industry should resist thinking in terms of "one size fits all" or "this is the universally best approach" when it comes to optical communications…

Stephen Hardy

In the feature article "Timing is Everything" that begins on page 23, Jim Theodoras of ADVA Optical Networking and the Ethernet Alliance reviews the different options carriers can pursue to provide timing in packet-based networks. Each option has its pros and cons, Jim asserts. So which one does he think operators should use in their networks? In all likelihood, more than one, he concludes. No alternative is appropriate for all applications.

We can extend Jim's logic to a variety of other scenarios - despite what we might believe the "right" approach should be. Jim's article takes as its premise the evolution towards packet-heavy traffic mixes. But "traffic mixes" is not a misnomer; many carrier networks will have to support services that, at least at the customer premises and perhaps well into the access network, are based on SONET/SDH or even PDH. The final demise of SONET/SDH will remain more of a "long heralded" rather than an actual phenomenon for several more years yet.

Similarly, while many of the discussions at conferences and trade shows has revolved around 40- and 100-Gbps capabilities, few networks will need this level of capacity in the next several years. Many connections have just reached 10 Gbps.

And while the flexibility tunable transceivers and reconfigurable optical add/drop multiplexers (ROADMs) offer has long been touted, such technology still doesn't appear in the lion's share of network applications.

Meanwhile, the fact that this month will see both the FTTH Conference in Orlando, FL, and the Broadband World Forum in Paris highlights another area where we should resist the temptation to assume that what some have proclaimed the future is truly a fait accompli. As I admitted in last issue's "First Take" video, I'd love to believe that fiber will reach most homes around the world - or even just in the United States - in the next five years or so. But that's just not going to happen. For every Verizon that has the vision and the financial wherewithal to replace its residential copper lines with fiber (or most of the ones they didn't sell, anyway), there are at least half a dozen service providers determined to squeeze every last kilobit out of their copper plant. Many carriers will even take the hybrid route Jim Theodoras describes in his network timing article, using fiber where it can and copper and perhaps wireless where it can't.

Network infrastructures generally just aren't pure IP, pure fiber, or pure anything. Despite our preferences, like life, networks aren't black and white. They're closer to plaid - different colors (or technologies) blended together in as coherent and complementary a way as network architects can manage. Most of the time the threads that make up that blend are so tightly bound that they prove impossible to tease apart.

So let's resist the notion that one technology will fit all applications. Networks in general may be moving to all-packet or all-fiber or all-whatever-you-want. But while they may be traveling in a discernable direction, it's unlikely they'll reach the end of whatever road you see them on.

More in Network Design