Ken, you may recall that we've discussed this topic on various threads in the past, including the Gilder and TLAB threads, going back to when dwdm's were first released. And then later again, with increased fervor, when lambda routing was first introduced.
I don't know what people mean when they say "the demise of Sonet," because it exists in more ways than simply the most obvious, e.g., the now-familiar self-healing rings and the add-drop capabilities in SONET muxes and digital cross-connects. The latter images of SONET are, for the most part, what the pundits are referring to when they talk about demise, although the latter two aspects are the electronic derivatives of the SONET model. Notwithstanding, one must keep in mind that add-drops and cross-connects are now supported in the optical domain, as well, and when they are they are usually conveying the same SONET-rated flows.
Stated another way, where wavelengths are add-dropped and cross-connected, instead of their electrical counterpart signals (STS-n's), they are labeled in accordance with their SONET framing rates such as OC-48, -192, etc. The same goes for the highest capacity routers in existence today, as well. I am waiting to see more native IEEE port speeds on those edge and core routers, but as of yet the vast majority of them continue to be SONET rated.
The article from Light Reading went rather deep in demonstrating SONET's staying power in the metro network venue. But even in the WAN, where (d)wdm is now responsible for supporting dozens, hundreds of wavelengths and in turn supporting MPLS and variants thereof, one must stop to examine how those wavelengths are being packaged.
With very few exceptions, they are containerized in SONET frames, typically rated as OC-48, OC-192, and some now beginning to dip their toes into OC-768, with at least two vendors now boasting that OC-3072 is just over the horizon. The exceptions to this rule lie in the support of GbE in the metro, primarily, and some instantiations of Fibre Channel where they are not being bundled in a SONET pipe.
But when you step up to 10GbE, it appears that at least one approach to porting this IEEE protocol across distances will be to fashion it into an OC192 payload (being 9.953 Gb/s, not 10 Gb/s), somehow (which will result in the stream being minus a few bits... that no one will miss, it's said...)?
If you wish to close one eye and grant the SONET slayers their wish, you can probably discount the optical flows coming out of the big-iron optical switches and routers, even though they are SONETized, i.e., they are rated in SONET denominations.
Overall, however, I'd say that the claim that SONET is dying is way premature. The claim itself, however, is equally harmless. Get used to it, you'll be hearing it for a long time.
What we will see more of over time is the elimination of several shims in the SONET stack that are no longer necessary. But the SONET containerization aspect will be with us for a long long while. Proof of this comes to us from the leading optical network element vendors who, without exception, continue to rate their highest speed ports in SONET and SDH terms.
I read somewhere recently that legacy technologies are called legacy because they don't die. Not very glitzy, granted, but right to the point.
FAC |
