Hello elmat,
I find myself discussing this topic almost incessantly lately with clients, ex-students and old collaborators alike, and here in SI. The fiber shortages are indeed real, as I am now finding out first hand from suppliers who are requesting six to nine months of lead time for *multimode*!, never mind singlemode. The latter is to a great part attributed to emerging GbE and 10 GbE over MMF, and the need for "Next Gen" MultiMode Fiber to fill these roles over distances longer than 88 meters, and a desire on the parts of end users to avoid using special mode conversion/offset patch cords that was necessary at the higher speeds when using "legacy" MMF.
The term "Next Gen" m-m fiber is a vendor marketing thing that has been picked up in some of the standards setting discussions, too, to differentiate it from legacy MMF. It is already being deployed today.
Which has the potential to create a surplus in inventory of the legacy stuff, while producing a demand crunch on the new.
I invite you to do a search on Corning's Infinicor MMF in this thread or at their site if you'd like more information on this type of MMF and why it is needed. LU has a cut of this too, but I forget for the moment the brand name they use for it. Anyone know?
But the larger fiber optic glut issues that are often discussed have to do with extant fiber and routes as much as they do with future ones.
As you dissect the issues you realize that the answer to this puzzle defies the obvious parameters discussed in the press. Which is to say, it goes beyond the millions of fiber mile claims, the routes covered, and the number of lambdas that are carried along any route. And it even goes beyond the haves and have nots issues. It comes down to more pertinent issues, such as connectivity between two parties who do have access to fiber providers' networks and wish to communicate with each other using the protocol or services of their own choice. But they find that they are subjected to the line formatting that is available to them, as dictated by who is serving their locale, or building, or floor (as it now turns out, with BLECs entering the picture).
For example, if a serving area has an abundance of ILEC fiber that speaks SONET only, and the end user wants to communicate over a 100 Mb/s Ethernet link, then all the fiber in the world provided by the ILEC (or the MSO for that matter, who usually limits their service group to TV and DOCSIS-compliant data) is useless to our user who wants to communicate over Ethernet, or Fibre Channel, or ESCON, etc. You can even take this scenario and apply it in reverse. So much for the known protocols. In order to access the new carrier's cloud that is LAN protocol based, the user often must take a much larger SONET pipe from the ILEC (OC3 for 100 Mb/s Ethernet, or an OC 48 for 1000 Gb/s Ethernet) and then do the necessary conversions in order to complete the loop. By that time, all of the excess bandwidth in those SONET pipes and conversions that are necessary to get to the desired protocol are more expensive than the potential savings that would have resulted from having gone with a native LAN protocol.
Now, take the type of wavelength swapping/conversion scheme that John C. introduced just a few posts back. Neither the ILEC nor the MSO, nor most other overbuilders and resellers could support this scheme on an end to end basis because it does not have reach, or a presence beyond the few participants in the trials taking place. Again, an abundance of fiber, but it is useless to those who wish to network network using a new protocol that depends on optical wavelength bypass, say.
So, is a glut a glut if it cannot be used by emerging technologies?... or even existing technologies that a particular service provider cannot or will not support? (Which usually results in the proliferation of other providers who "will" support those protocols!)
This is only one aspect that I have been examining. Another aspect has to do with the interoperability of different "grades" of fiber. One fiber carrier who I routinely do business with on behalf of my clients is now stuck with having to issue several different levels of SLA to the same customer, depending on when a particular site went on line. While this may not seem like an issue for field offices, it becomes quite apparent to those who must manage a central site, where all grades of fiber are present.
This is because of the different grades of fiber that were used (each representing a new "generation"), which could be viewed as the rings of a tree, telling just how old an installation is by the type of fiber that was used. The most poignant thing to keep in mind here is that all grades of fiber don't all talk to one another in the same way, and some of them not at all, or very poorly and only over short distances, if the wrong wavelengths are used.
And still another aspect of the glut puzzle is the distributed nature of all fiber, being put into the ground along the same routes, often, but terminating in different carrier hotels and end points by carriers who don't support the same protocols, often. The implications of this become clear when you examine where these same carriers "don't" cover the same routes, out towards the edge and into communities. Now, if you want to get from corporate location A (say HQ) to corporate location B (a field branch office, say) you often have to use multiple carriers (which is often dependent on who they are peering with or with whom they have hand off relationships), and in so doing you must traverse twice as many nodes. Here, I use the term nodes to connote carreir hotels or central offices. Twice as many nodes as you ordinarily would if you were using only a single carrier. This mitigates some of the glut concerns, even though some of the bandwidth is being used only because there is no other way to get from A to B.
Oh, and each time you use another CH or CO you must use a subscriber section of cable to get into and out of each node. Tail sections add up rapidly, and worse, almost always the only way to get these is through the ILEC through SONETized connections. Which means that if you are doing 100 ro 1000 Mb/s Ethernet you must do conversions from SONET to Ethernet at extra costs.
Okay, so we go to an all Ethernet environment, eventually, right? What does that do for NGI protocols, a la the one introduced by John C upstream and the many others that depend on end to end opaqueness, with forward looking self-healing and problem avoidance algorithms? At that point Ethernet will be the next legacy protocol to overcome. And so it goes...
I'm sure I haven't answered your question adequately. Nor have I answered my own to my own satisfaction. Let's hear from others. While this may appear to be an old topic, and one that we have hacked to death, we should keep in mind that the underlying dynamics that go to evaluating it keep changing. All fiber is not equal. Nor do all carriers treat fiber in the same way. And the longer the generational gaps are that exist between fibers that have already been placed in the ground and the newer ones going in at any point in time, the less likely it will be that they could be joined together to produce end-to-endness. Which suggests that to a great extent we will have tubes in the ground that will house inferior or non-compatible fiber, as time goes by. And such fibers cannot be considered a part of a glut if they are useless for emerging protocols, and cannot be interconnected with newer grades that are being placed. This off-the-top-of-the-head message wasn't very eloquent, but it states some of my thoughts on the subject.
FAC |
