Mike, Denver, and Thread,
My previous comments re: the nature of the optical cross connect's [OCC's] interfaces were intended to highlight the fact that increasingly, optical layer access is going to be a key feature attribute with large users, and carriers, alike. When users have a need to deploy IP or ATM directly over lambda, or when they will soon have a need to break down optical spectrum in order to perform new forms of optical routing and path switching, SONET and SDH tend only to get in the way of things. They beg for numerous back to back to back to... conversions.
[Before I forget, Mike, the majority of Cross Connects are probably owned by ILECs, but every form of major league carrier uses them. My point wasn't that the ILEC would control the fiber exclusively, rather anyone who owned and operated an OCC, or a DSC for that matter, would be in that position. It had nothing to do with ILECs, specifically, in other words.]
Today's DWDMs yield an arbitrary number of lambdas, or wavelengths, whether it's 16 or 32 or 40, or whatever, with each lambda possessing a width of approximately 50 GHz or 100 GHz, depending on the optical channel plan in use.
My belief is that in the very near future large users will want access to an entire lambda, or multiple lambdas out of a possible much higher number per strand. The trend will soon be seen where the carrier will be leasing out lambdas as they now do OC-12s or OC-48s, or whatever denomination of SONET or SDH one could think of.
In turn, the user will want to be able to subdivide the lambda into smaller parts through the use of beam splitters, or lower order WDMs, as it were. Think of the possibilities that exist in enterprise networks that have this kind of spectrum at their disposal.
This is not far fetched, since users are already using two and four lambda devices to support IBM SNA ESCON connections on one or two lambdas, while conducting business as usual on the other two. Or, they may be using one or two for ESCON, one or two for GE, and one or two for traditional WAN protocols. But for the most part they are still using their own native glass to do this at this time, or in rare instances they are obtaining dark fiber from carriers, like MFNX. Things are about to change in this respect, I fully believe.
------
The point is, that large users, and carriers alike, don't like to have their options taken away from them simply for the sake of conforming to an international standard, especially when it's going to cost them a bundle to convert the OC-n to IEEE or to an ANSI standard, and then back again, perhaps several times, anyway.
So, when I suggested that the models that we've been looking at at the present time (both the Tellium and the TLAB) only have SONET or SDH interfaces, I really meant to say that they should also be outfitted with an assortment of pure lambda interface capabilities, as well as perhaps IEEE and ANSI interface standards, like Fiber Channel and FDDI, at the lambda level, so that users and carriers alike could use them in a native manner, keyed to their requirements.
Incidentally, there already exist several devices (check out MRVC's line, as an example) that already conform to these principles, although they don't conform to the larger OCC parameters, in that they don't do DACSing, or DCS functions at the central office level, per se.
Regards, Frank Coluccio |
