why wasn't one of the options "GMST is a miserable piece of <deleted> that is destined to wreck my unborn child's ultimate prospects in life"?
less seriously, inspired by that incredibly simple recent discussion of storage, I thought threadsters might be interested in a much more arcane treatment of photonics--the first in a series from a savvy Briefing.com analyst (same guy who did the B2B series that godzilla threadsters might remember).
tekboy/Ares@Q-2KinY2K.org
Photonic Revolution: Part I
[BRIEFING.COM - Gregory A. Jones] After finishing up a B2B series late last year, we promised a future series on the Internet backbone sector. A millenium later, it finally gets underway. But the focus has narrowed somewhat. Though this series will cover the backbone, it will focus primarily on the fiber optic companies that are the future of the telecommunications infrastructure. To open the series, we will offer a broad view of the fiber landscape, after which we will look in greater detail at specific sub-sectors.
Why Photons?
Before getting to the forest, we need to discuss the tree. In the fiber optic network, that means the photon. Electrons travelling through wires were the old network. Photons travelling through glass are the new network. The reason is simple -- photons travel through glass much more quickly than electrons travel through wire. In short, optical delivery offers far more bandwidth potential than electrical delivery.
The Analogy
Understanding any one piece of the Internet backbone is a relatively easy task, but putting all of the pieces together into a comprehensible whole is another story. This article will therefore strive to achieve that one lofty but focussed goal.
Our own photonic revelation was that the fiber optic network bears great resemblance to the transportation network. Not data or voice transportation, but a network we can all understand -- interstates, highways, roads, and alleys. So that's where we will begin our optical journey.
The Interstate
The core of the telecommunications network is long strands of fiber. The core of the transportation network is long strands of interstate highway. Both have a simple function -- transport a lot of traffic in a straight line. In both cases, this is a relatively easy task. You don't have to worry about getting a packet of data to a specific IP address or the AMC Pacer to a specific street address. You just have to get a bunch of traffic from East to West, North to South, etc. So you build fat pipes or wide, straight roads and get the job done. In both cases, this part of the network does not demand much intelligence -- just big bandwidth.
The Cloverleaf
Just as no one's driveway is accessed directly from the Interstate, no one's Internet connection taps directly into the backbone. So in both cases, we need the ability to redirect traffic to get home. On the Interstate, that is done with massive cloverleafs -- networks of overpasses that get the Southbound traffic heading East, etc. These cloverleafs still don't have much intelligence -- they don't get you home, but they do redirect you to get closer.
The telecommunications equivalent is a huge switch, also known as a wide area cross-connect. Current cross-connects are electrical. But the bandwidth capabilities of electrical cross-connects are not keeping pace with the bandwidth capacity of the optical core. DWDM (dense wave division multiplexing) technology has increased fiber capacity by 40 fold in current systems, and perhaps by several thousand fold in future systems. Imagine doing the same with an interstate but not doing anything to the cloverleafs. You could deliver huge amounts of cars down the interstate, but they would pile up at every interchange, as 40 lanes of interstate dump onto a one lane cloverleaf.
Such is the case with today's electrical cross-connects. The solution is the all-optical switch, which promises the potential to route traffic at the speed of the core. It is the equivalent of expanding that cloverleaf to 40, 80, or however many lanes the interstate has. It's not possible in the transportation network, but it will be a reality in the optical network.
Highways, Biways, and Beltways
When a car finally leaves the Interstate to get closer to its final destination, it ends up on various metropolitan roads which can sometimes be as large as interstates, but which have to get the traffic a bit closer to home and face more difficulties in dealing with intersections as they can be interrupted by additional exits, onramps, and sometimes even traffic lights -- all the necessary evils of getting you to your final destination.
In the communications network, we have the metropolitan core -- usually fiber, but with less bandwidth than the core network due to the high cost of implementing DWDM relative to the potential benefits. As with wide area cross-connects, these metro networks can be another bottleneck. We have massive bandwidth at the core where DWDM is an obvious benefit meeting more limited bandwidth in the metro network where DWDM doesn't offer the flexibility needed to deliver a wide range of bandwidth and service to the end-user.
Overpasses, Underpasses, and Eight Lane Intersections
Within the metropolitan transportation network, intersections become much more varied than the cloverleaf, and designed for a much broader range of traffic flow. You have overpasses, stoplights, stop signs, etc. The goal is once again obvious -- get the traffic to its destination.
In the communications network, this is the edge. The edge is where high bandwidth optical traffic must be broken down into myriad revenue-generating services that RBOCs and CLECs sell to businesses and consumers. This is where bandwidth is still critical but intelligence is even more important. Metropolitan area access equipment is the product at this point of the network, and is having somewhat better luck in keeping pace with bandwidth, but only because prior bottlenecks in the network have reduced metro traffic to the point where today's electrical switches and opto-electrical hybrids are generally up to the task. But as DWDM makes inroads into the MAN (metropolitan area network), a bandwidth mismatch will arise here as well and optical solutions will win.
Roads, Avenues, and Alleys
Now we're on the 25-mile hour Elm St or the pothole-laden alley. In the telecom world, this is the end of fiber and beginning of copper wire. This is the last mile, and in both worlds, it's slow going. You can't dump an interstate into your driveway, and you can't pull an OC-192 (10 gbps) fiber strand into your living room. You are stuck with 56K copper wire at worst, or hopefully DSL, cable, or broadband wireless. And maybe in the future, fiberless optics. In any case, this is still the weakest link in the chain, and the point at which much of the advantages of booming bandwidth at the core are wiped out due to narrow bandwidth in the last mile.
The Driveway
Home at last. Whether it's the car pulling into the driveway, or circuit or IP packet finding its final destination at an office or residence or colocation facility, this is the end of the road. At this point, fiber is very rare with the exception of the most bandwidth hungry of enterprises. Cat 5 LAN cables and gigabit ethernet can handle all but the most bandwidth-intensive applications. And it is here that sales of all manner of switches -- layer 3, layer 4-7 -- are selling like hotcakes (most of them by Cisco). Or in the consumer realm, it is an analog modem, DSL modem, cable modem, or wireless network.
The Road Ahead
This has obviously been a very broad and vague overview of the fiber optic landscape -- a whetting of the appetite. In future installments, we will start with the core of the network and move outward, discussing the trends in fiber technology, and the public and private companies that stand to benefit or get trampled by these trends. We'll start with the technology that has led to the bandwidth explosion: DWDM. |
