I'll have to agree with Ray, Chris.. that was a good find, affording plenty of room for debate both here and elsewhere... as some of the individuals in the article itself have demonstrated.
At this rate I'll find it very difficult to catch up to some of my old messages... but I couldn't resist this one because it strikes home.
I don't know if I agree that Layer One Sonet is designed to be more supportive of switching rather than routing. It's a physical layer component of the network and cares not what rides atop it, I would think. I think that I would need to hear more from the author as to why exactly he stated this. Unless, of course, he meant that the frame size was an extension of multiples of voice circuits, which are themselves traceable to the digital-equivalent of coded PCM words which represent the audible range of the human ear... nah.
Sonet will begin its demise in the long-haul all-IP backbones, first, where the number of I/O ports per mile are minimal, and the efficiencies afforded by DWDM are maximal where it counts: On the fiber routes.
Sonet will persist in the distribution plant to local end offices and to large enterprises, however, since port provisions to end users are already in place and these are not as economically supplanted in the last mile as they are in the long haul. Far from it now, but price convergence has begun to take place to offset some of the costs already, especially in the denser business districts.
The jitter problem? If you don't clean it up in the backbone you must compensate for it at the end points in software. Sooner or later, you pay the piper.
One must be careful, however, when using the term "jitter." This is because jitter has several different connotations in networking, most notably: (1) in the time domain, in an instantaneous sense [peak to peak jitter of pulse rise- and fall- times, i.e., a measure of the subtle differences in periodicity from one pulse (or sine wave) to the next], and (2), in the delivery times of sequenced packets (inconsistency of arrival times and the artifacts of differential delays upon the received message, most notable in video and voice over IP payloads).
I happen to agree with Eugene Park, director of strategic marketing at optical newcomer LuxN Inc.:
"...digital wrappers are just Sonet all over again."
The wrappers are made up of many of the same overhead bits that exist at the beginning of "regular" SONET frames. What's missing in the streamlined versions, in additions to some of the bits (which are stuffed as null bits) is most of the underlying administration and traffic management hooks, along with some of the self-healing mechanisms, but the framing intervals and packaging sizes remain the same.
You might be interested to know that some of the remaining companies mentioned in the article have been discussed at various times here in SI on the SilkRoad thread.
Regards, Frank Coluccio |
