JRH,
The means to achieving higher bit rates, as in terabit rates, is in stacking multiple OC-192s (or other high bit rate streams) on top of one another in a strand, by using multiple wavelengths, as in DWDM. Seldom is the intent to actually derive a single feed of those transfer rates, rather, the vendors are talking about aggregate rates of multiple streams. There's a lot of marketechture (smoke and mirrors) in this game, as well. If we look at the OC-192 feed itself, it is, in itself, an aggregation of many lower speed tributaries, right down to the basic building blocks of bytes, DS0s (64 kbps, or 8000 bytes) and T-1s (1.544 Mbps).
I think (not sure at this point) that the highest level of concatenation to date is an ATM envelope in an OC-48c, or 2.5 Gbps, if I'm not mistaken, at least in the commercial sphere, that is.
Many of the Terabit announcements in the past year have been fooey, only leveraging the new-found capabilities of WDM/DWDM, and using this form of compounding the many individual lower speed tributaries, all along. Previously, the same speeds were achievable over separate optical fibers. What's new about the emerging stuff, is that they are using optical layering, a form of frequency division multiplexing like coaxial cable television techniques, within the same strand.
There are, of course, many other benefits being derived from DWDM and new methods of photonic manipulation, but these are not addressed, by and large, in the same breath as the sensationalizing statements associated with aggregate flow rates. Much more important than aggregation, IMO, are the new routing and add-drop capabilities we now have to work with. Speed is extremely important, too, but ease and flexibilities in administering the optical flows is where some of the real gains come from.
Another method could actually take into account a much higher single tributary data rate, or a higher master flow, but bus speeds and processor speeds are the bottlenecks to achieving this at this time. The current cost-benefit is in favor of using WDM, and will probably remain that way for a long while, or at least that is the belief of many pundits in this area.
If we use my cable tv example to make a point, you might have a 900 MHz Hybrid Fiber/Coax (HFC) system feeding your neighborhood, but at the individual residence level, you only use a single 6 MHz channel at any one time. See my point?
If anyone knows of a higher-speed single data stream than OC-48c, I'd like to know, too.
Regards, Frank Coluccio |
