Degrees of Twaddle

"How long is a piece of string?"

Optical networking vendors don't know the answer to that old chestnut any more than you or I (Steve Saunders claims to know the answer, incidentally, but that's just Steve being Steve).

But based on the nonsense they pulled at the NFOEC conference and exhibition last week (in Baltimore, tragically) vendors would surely have an opinion on the correct unit of stringometry ("A cord!" "No, a spool!" "Fool! It's a fur-long!" "You stinking liar...!" and so on).

Basically, when it comes to how to measure different aspects of optical networking, vendors at the show used more benchmarks than a home improvement show.

This isn't just a case of engineers with too many brands of slide-rules in their pocket-protectors, incidentally. A lot of the time a vendor's decision to use one unit of measurement over another is based more on business and marketing considerations than engineering preference. For instance, choosing a really obscure, olde worlde unit of measurement can be a great way to impress investors – and ensure that no-one can work out what your company is really up to (anyone for cubits?).

I first started noticing vendors' measurement mores when Optical Switch Corp. announced its new phase mask developments aimed at making FBGs (fiber Bragg gratings) for chromatic dispersion compensators (see Optical Switch Announces Components).

As it happened, Southampton Photonics Inc. also announced FBG developments aimed at exactly the same application at the same time (see Southampton Ships FBGs, Names VP). And that’s where the fun began.

Optical Switch said its developments could make chirped FBGs (cluck cluck!) with a resolution of 13 nanometers per centimeter, while Southampton Photonics made a big thing of its group delay of plus or minus 3 picoseconds. To be frank, I was in a total fog. How did the two compare? After a lot of heavy number crunching, Optical Switch came to the conclusion that the end result was broadly similar dispersion compensation performance (in other words, who cares?).

Anyhow, that got me started, so I asked other folk at the NFOEC how they measured things. Professor David M. Bloom, a founder of LightConnect Inc. (see LightConnect Comes Into Bloom) came up with a couple of classics.

In the early days of lasers, Bloom says, their power was measured in “Gillettes” – the number of razor blades they could burn through.

Bloom also put me onto A Dictionary of Units of Measurement compiled by Russ Rowlett at The University of North Carolina. It cites some really strange units, including “gillions”, an obscure way of saying thousands of millions (giga-millions) for purist Brits, who think billions should refer to millions of millions. The best one, however, is “degrees Twaddle”, a measurement of specific gravity of liquids denser than water.

This gives me a suitable segue to bring up the degrees of twaddle surrounding another unit of measurement at the moment – the percentage utilization of fiber capacity – which got a big airing at the NFOEC, courtesy of Corvis Corp. (Nasdaq: CORV) and Williams Communications Group (NYSE: WCG) (see Williams, Corvis Lash Back).

In essence, analysts have come up with the notion that having lots of fiber in the ground and using very little of it is bad. This penalizes new carriers unfairly because it hardly costs them any more to install cables with, say 144 rather than 24 fibers in them. It also fails to recognize that modern fiber can support far more DWDM channels than older fiber, giving new carriers an intrinsic advantage over established ones, according to David Smith, VP of hardware engineering at Corvis. On top of that, new carriers may be just getting started on filling their infrastructure with traffic, he notes. See Fiber Utilization Figures Challenged for more on this subject.

Another type of twaddle was evident at a breakfast meeting held by Altamar Networks at NFOEC that I attended. My colleagues at the breakfast reckoned that the price of telecom equipment is nearly always reflected in its size. By that reckoning, someone joked, vendors could sell their gear for so much a pound, like potatoes.

Plenty of other optical networking phenomena are measured in units that non-physicists have trouble getting their heads around – like decibels, which don’t appear to follow a linear scale, and bandwidth measured in GHz rather than Gbit/s. Happily, Light Reading has just published a Beginner’s Guide on this subject: Optical Units Reference.

If you know of some other examples of strange ways of measuring optical phenomena, please share them on the message board following this article.

— Peter Heywood, Founding Editor, Light Reading
gea 12/4/2012 | 8:04:07 PM
re: Degrees of Twaddle I haven't seen this so much at NFOEC, but it's rampant at OFC: the throughput of an optical cross connect measured in bits/second.
This gets silly quickly when applied to MEMs-based OXCs: any MEMs mirror can actually reflect potentially many wavelengths at a time, so the more wavelengths you can generate, the higher the "bit rate" of an OXC.
One day I want to give a talk where I claim to have an OXC that can cross connect 100Tb/s, and then I'll say "and it fits in my pocket!"
Then I'll pull out a small section of optical fiber, and explain that the "fiber cross-connects hundreds of 40Gb/s wavelengths from this port to that port".
fiber_r_us 12/4/2012 | 8:04:02 PM
re: Degrees of Twaddle Got to agree with you here gea... The whole "my OXC is a massive router that passes 100Tb/s" mantra is about the most ignorant marketing I have seen in a long time. MEMs-based switches are fancy patch-panels. They have no inherent ability to deal with data in terms of "bits-per-second".

The term makes a little more sense when you are talking about OEO switches (i.e. broadband crossconnects) such as CoreDirector.

I think it makes the most sense to talk about bit-per-second throughputs in the packet switching world.

There is no logical reason why you would compare bit-per-second through a TDM-based crossconnect and something like a packet router. These are two separate and complimentary technologies that have no need to be compared in terms of bits-per-second.
ownstock 12/4/2012 | 8:03:15 PM
re: Degrees of Twaddle You are spot-on, dude. This is where you people at LR should be all over the marketing idiots at the various companies.

As the earlier post said, fiber optics ain't no big deal: basically hoses with (OK, colored) water in 'em, batch panels, etc. So for switch companies, the questions are 1) n x m (fill in the blanks) and 2) how much?

Most of the market really is that simple...but if anyone admitted it, and changed their behavior, their recent investors might figure it out too...oops! And potential customers might not call to find out if there really was anything there, and then they would never hear from them...again, oops!

So, the message is, if you smell twaddle cooking, better run like hell while you still have a rear hip...let alone pocket...let alone wallet!
Peter Heywood 12/4/2012 | 8:03:11 PM
re: Degrees of Twaddle Ownstock, thanks. I've been promising to do a big thing on optical switches since ...well, several months ago. When I get a momemt (ho ho) I promise I'll put all of this down in black and white.
tiredofit 12/4/2012 | 8:03:05 PM
re: Degrees of Twaddle Can somebody quantify how much capacity can be supported by a shitload of fiber? :-)
lightmaster 12/4/2012 | 8:01:11 PM
re: Degrees of Twaddle This one is my favorite. Backplanes capacity is rarely if ever a factor in system performance, unless the system design is totally screwed up. It's like talking about the theoretical capacity of the physical optical fiber.

One MSPP company even made an announcement of their backplane capacity last year. Their system was bound by I/O and switching capacity to a small fraction of the theoretical capacity.
Peter Heywood 12/4/2012 | 8:00:55 PM
re: Degrees of Twaddle Which MSPP company was that? I feel like embarrassing somebody.
tiredofit 12/4/2012 | 8:00:30 PM
re: Degrees of Twaddle Gotta love 'objective' reporting!!
lightmaster 12/4/2012 | 8:00:29 PM
re: Degrees of Twaddle Peter,


It's not a bad company at all, just really bad marketing.
Sign In