Friday, July 11, 2008

FTTD (Fiber To The Desk)

I've long thought that FTTH (Fiber To The Home) was the last step in high-speed networking. We've already got easy access to 1 gig ethernet for in-home use. After all when we get a 100 meg symmetrical intranet here in Lafayette we'll be far, far ahead of any connection available to large numbers of people in the US. Most folks dream of a 2o meg connection—and even more dream of being able to afford it. So the 1 gig ethernet network in my home seem, by an order of magnitude, far more than adequate to handle my practical needs. Fretting about an in-house fiber network seemed plain silly.

But maybe FTTD (Fiber To The Desk) won't seem silly for much longer. According to a recent NetworkWorld article Verizon is actively marketing and has already installed a few FTTD setups in larger commercial and institutional settings. According to the article FTTD has real, solid advantages right now:
  • cheaper materials (copper is costy),
  • cheaper to install,
  • faster installation,
  • fewer electronics,
  • cheaper to maintain,
  • space-saving (fewer maintenaince closets),
  • energy-saving.
  • and oh yeah: far faster now and future speed upgrades will be dead easy
Now that's a pretty compelling list. The article is very straightforward, however, in saying that FTTD isn't currently for everyone. You need to have a fairly large population of users and retrofitting a smaller building probably wouldn't be cost-effective. And my own caveat: in reading between the lines of the story it is pretty clear that the cost savings are based, in part, on eliminating the need for a separate phone network. The user is assumed to be on a VOIP phone. Now that seems like a sensible long-term evolution but it may mean that you'd come up with different numbers if you did a straight ethernet to fiber comparison. Still...even being in the same ballpark is very surprising and since phone wiring is going the way of the dodo anyway it may make sense to disregard its cost.

So FTTD is not for everyone...Yet.

I (cough, cough) remember when ethernet was viewed as similarly impractical for home use. It was an enterprise or institutional networking infrastructure. Nobody would have troubled themselves to install it in a household or small business. —I, back in the dark ages, installed AppleTalk networks in a graphics firm and in the building where I did graduate study. Ethernet was thought to be overkill--useful in the backbone but too much and too technical for mere mortals or small groups.

As to those who can't imagine that we'll need that sort of bandwidth the same ethernet history applies: There was a time when a powerful local network seemed pointless. After all what household could afford more than one computer and if you had two, why would they talk to each other? But as of this morning I've got a total of 6 or 7 computers pulling power in my two person household: 2 main laptops, 2 TiVos (each of which is a linux computer) and 2 "extra" computers--one serving as a "kids or visitors" laptop with appropriate programs installed and one running as a server/workstation. If you wanted to count the Cox set top box that would make 7. And hey, I don't have an AppleTV, iphone, videophone, xbox, PS2, or a Wii. (Poor pitiful me!) I do have a network storage device backing up the main computers regularly, so that eats some bandwidth when it kicks in. A household with a lot of gamers or heavy video watchers could generate a lot more internal traffic than I do.

So when the claim is made that a networking infrastructure is more capable and cheaper for larger installations—and with local network demand for are visibly growing—I strongly suspect that it will, in short order, be feasible and valuable for home use. My guess is that somewhere shortly after we see 10 gigs being pushed to the wall of our homes we'll think our 1 gig of in-house ethernet is puny. We'll then think it makes sense to disconnect our ethernet, lash fiber to the ethernet wire and pull the copper ethernet out as we run the fiber in behind it.

Digging Deeper....
For those who might be interested in a bit more technical background (as far as I understand it): FTTD as discussed above is really a form PON (Passive Optical Networking). Verizon who is featured in the story uses this form of networking in its FTTH network FIOS—LUS will use the same framework here in Lafayette. PON systems take a single fiber feed and split the bandwidth it carries between a number of users; in most metropolitan-level networks it is split between 32 users and it appears that the Verizon commercial product discussed is based on this model. So there would be one "cabinet" in a large building that would replace the central IT hub and various closets with electronics like routers that are typically found on every floor or every large segment of an ethernet network. Part of the savings is found there. From the hub fiber would run to every desktop and that would necessitate an ONT (Optical Network Terminal) at every desk to translate the fiber's light to electrical signals for use in the computer, TV, phone, or other device (like a wireless node, network storage, or network printer). The cost of this ONT would be a major issue, I would think. I would expect that some day the ONT would be integrated into the device itself as has happened with ethernet and other networking technologies.

So there really does appear to be a path to FTTD @ home in the foreseeable future. All we need is enough bandwidth to our houses to make deploying the in-house sensible.


FAC said...

Very interesting observations, and all of them spot on. You'll be interested to know that plastic fiber optics (POF) are now suitable for transporting in-home v/v/d at 100 Mbps at distances of 100 meters, and 1 G at distances approximating 40 meters at the present time. This is a far cry from the 10 meters that POF was capable of handling only a decade or so ago at much lower throughput rates.

POF is also very simple to splice and connectorize, taking only seconds, where earlier tasks could take anywhere from fifteen minutes to a half-hour using silica-based fibers, associated connectors and related termination hardware.

Incidentally, as I pointed out in my response to the Verizon-SAIC article you referenced ("It's about time ..."), PONs have been around for a very long time. The ITU's FSAN reference model during the Nineties, and now the IEEE's Ethernet in the First Mile specifications, came along more recently, but PONs using CSMA/CD protocols were used as far back as twenty-five years ago using very simple optical power splitters.

Your observation concerning the still excessively high costs associated with ONTs/ONUs is accurate, but those devices, too, when the bloat of "triple play" functions are removed from them for the purpose of supporting only standard Ethernet desktops, will yield smaller footprints and form factors, and use far less power-demanding electronics as well.

Finally, your main thesis holds up as well. What generally starts in the center of the network usually finds its way to the edges. Interesting post. Thanks.


John said...

Thanks Frank,

Interesting comments...I had heard a bit about optical plastics but have thought that the new bendable glass might eliminate its niche.


FAC said...

Hi John.

The 'niche' for POF has traditionally been industrial and automotive systems (under the hood of cars, fuselages of airplanes, and other machinery) where it satisfies myriad requirements economically without introducing or being affected by RFI/EMI. It's only been recent, in relative terms, that plastic has begun to attract serious attention in the LAN space, following improvements in distances and much higher throughput handling capabilities.

I'd submit that, beyond its bend-resistance, it is the ease of handling and implementation and its broader economic advantage that are most responsible for its newfound attractiveness. See my introductory note in brackets [FAC: ... ] and the article following it, here:


FAC said...

A new book on the subject of Plastic Optical Fiber, or POF, may interest you, although its price tag is a bit steep for casual reading purposes ;)

Book Release: POF Handbook - Optical Short Range Transmission Systems

Release: April 18, 2008; Price $274.95 (with FedEx Shipping -- geesh, I should hope so!)

Authors: by Olaf Ziemann, Juergen Krauser, Peter E. Zamzow, and Werner Daum

This recent post of mine, too, which I had attempted to post here earlier, but for some reason has yet to appear:

FAC said...

John, as an additional FYI on plastic optical fiber, see: