Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Ciena (CIEN) -- Ignore unavailable to you. Want to Upgrade?

To: Gary Korn who wrote (1290)	2/20/1998 1:47:00 PM
From: Intel Trader	Respond to of 12623

Salomon upgrades CIEN. it

To: Gary Korn who wrote (1290)	2/20/1998 2:32:00 PM
From: Maverick	Respond to of 12623

Gary, I've been involved w/ UDWDM technology lately and fairly intimate w/ some key players in this field. Per your questions, 1, 2, 3 are true. However, the high PE (53.33) commensurates to its PE growth rate. Let's check out its competitors in my next post.

To: Gary Korn who wrote (1290)	2/20/1998 2:36:00 PM
From: Maverick	Respond to of 12623

WDM: North American Deployment Trends By John P. Ryan, ryan hankin kent Three years ago, in early 1995, dense wavelength-division multiplexing (WDM) was still in its infancy, not in widespread commercial implementation, and the major network systems vendors were offering few WDM products; optical switches and all-optical networks were little more than the subjects of futuristic papers at expert conferences. In late 1995 and throughout 1996 and 1997 this all changed dramatically, as WDM became a billion-dollar mainstream business, creating perhaps the most rapid emergence of a new technology ever: Ciena's success at launching its WDM systems into use at Sprint and Worldcom led to an immensely successful initial public offering; indeed, the market valuation of the firm at the time of its first trade in February 1997 made it the largest startup IPO in history, and its first-year revenue total of $196 million gave the firm the fastest revenue track in corporate history. Lucent, which started shipping eight-wavelength WDM systems in 1995, says that well over 1000 of its WDM terminals are now in deployment. Every major U.S. long distance service provider was utilizing WDM systems as a standard part of their network by the end of 1997, and every major vendor of lightwave transmission systems was supplying the products. Sales of WDM transmission systems to North American network operators have soared from perhaps $50 million in 1995 to over $1 billion in 1997. Our firm's analyses forecast this market continuing its heady growth, to exceed $4 billion by 2001. Thus, in a few years the new technology has created revenues that start to rival those of established synchronous optical network (SONET) systems. A plethora of vendors have responded to the lure of this multibillion-dollar bonanza, with WDM products either delivered or promised from telecom industry stalwarts worldwide: ADC, Alcatel, Bosch, DSC, Ericsson, Fujitsu, Hitachi, Lucent, NEC, Nortel, Scientific-Atlanta, and Tellabs. But the new technology has brought its crop of startups: Pirelli's debut in the systems business, Ciena, Cambrian (an offshoot of Newbridge Networks), and Tellium (started by engineers from Bellcore). IBM's WDM products used to link campuses across leased dark fibers are now joined by competition from Germany's ADVA and Osicom.

To: Gary Korn who wrote (1290)	2/20/1998 2:38:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part II Optical add/drop multiplexers and optical switches are now on trial in many locations, with carriers experimenting on alternative architectures, and widespread utilization expected in 1999. This article answers the following questions: What happened to drive this explosion of optical networking technology in the North American market? What suggests that it will continue? What happens now to make the transition from WDM as a point-to-point network technology to real optical networks? The Drivers for This Explosion of WDM At a simple level, the principle driver for this explosion was an unexpectedly rapid exhaustion of the capacity of long distance fiber networks. This fiber exhaust, combined with favorable economics for WDM, led to the use of the technology instead of other alternatives. The sudden nature of the emergence of WDM derives from the accelerating growth of traffic and the sudden need to create a more open ended approach to creating capacity. For many years, there has been thought to be a fiber glut -- vast reserves of unused fibers. George Gilder has written (in Forbes ASAP in 1992 and again in 1997) about the telcos' reserves of unlit fiber. While the regional Bell operating companies (RBOCs) may have reserves of unlit fiber, it was increasingly apparent by the end of 1995 that the glut was gone in the interexchange carriers' networks. Indeed, the fiber backbones of the long distance carriers were nearing exhaust; and the rapidly accelerating pace of traffic growth meant that the major networks would be running at capacity in short order. The problem has its origins in how the long-distance networks were created: rapidly, in the aftermath of the long-ago 1984 divestiture of AT&T. At that time, the newcomers to the long distance market (MCI, the network that became Sprint, and others) were sufficiently cash-constrained that their networks were based on rather small fiber cross-sections. A typical fiber route might have just 16 fibers; fiber counts of 32 or more were relatively rare. However, back then traffic was mainly voice, and the overall load grew slowly.

To: Gary Korn who wrote (1290)	2/20/1998 2:39:00 PM
From: Maverick	Read Replies (1) \| Respond to of 12623

WDM: NA, part III But the 1990s are different. Traffic is increasingly dominated by data, not voice. Pacific Bell has reported that its traffic became predominantly data in 1995; all carriers report growing amounts of data traffic. Goldman Sachs' Mary Henry projects voice traffic continuing its traditional lethargic growth pattern, while video and data traffic continue an inexorable exponential growth, reaching as much as 70 percent of traffic by the year 2000. Specifically: voice traffic is estimated to grow at 8 percent/ year, data traffic in total (including modem dial-up access to the Internet and corporate frame relay, among other things) is rising at rates estimated at 35 percent/year. Internet traffic within that continues its dizzying growth rates of greater than 100 percent/year, creating the prospect that overall traffic growth in North America might actually accelerate further! Most important, telecommunications markets seem to have established a strong positive elasticity; in engineering terms, they have found a new positive feedback cycle. What this means is that if the unit price of a telecommunications service (e.g., 1 min of transmission at 1.5 Mb/s over 1000 km) declines by x percent, traffic will rise by (x + y) percent, so revenues will rise by y percent. Thus, for the foreseeable future, so long as unit prices for telecommunications services can decline, traffic will soar (and telecom revenues will rise). The realities of the competitive market bear this theoretical picture out: operators strive to capture enterprise traffic from each other and from older forms (e.g., leased lines) by deploying new technologies that lower the unit cost and price of their services. The results, in general, are rapidly rising traffic and higher network utilization. As if this rising tide of traffic were not enough, the major long distance carriers were in the process of implementing SONET ring technologies. SONET rings provide an important means to eliminate the risk of network outages caused by cable cuts. They use network systems that sense the failure of a link, and then reroute the signals, going in the opposite direction around the fiber ring. These systems have a fast restoration mechanism that is of great value in ensuring network reliability, but do so by sacrificing fully 50 percent of the fiber capacity!

To: Gary Korn who wrote (1290)	2/20/1998 2:41:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part IV Table 1, based on Federal Communications Commission (FCC) statistics, summarizes the percentages of fibers lit in major long distance networks. The table, already indicating the dearth of capacity in long distance networks, to a significant degree understates the problem, for some of each carrier's network may be available, but in the wrong place. The unlit fibers may be in rural areas, when the need is in urban areas. In fact, several carriers have told us that they consider large parts of their networks to be fully lit: absent a major upgrade path, there is no room for growth in traffic at all. And, at a time when traffic was growing at historic record rates, and the telecommunications infrastructure was becoming more important to enterprise end users, the giant operators were starting to have to tell their best customers that there was no more capacity left. That being the case, the stage was set for WDM to compete alongside the other upgrade mechanisms as the carriers found in 1995 and 1996 that their need for capacity upgrades was becoming desperate. Meanwhile, the new technology's capabilities were rising. WDM Reaches Maturity At the same time, the underpinnings of WDM technology were maturing. Two-wavelength WDM has been in existence for a decade or more, combining transmission at 1310 nm (the wavelength where traditional single-mode fiber has zero dispersion) with transmission near 1530 nm (the wavelength region where silica-based single-mode fibers have their lowest attenuation). However, as a two-wavelength technology, the capacity of a fiber could "only" double, and with span engineering quite different for the two wavelengths (span lengths commonly would be attenuation-limited for one but dispersion-limited for the other) this form of WDM had become not much more than a device-level "gimmick," an upgrade of last resort, but not the basis for network planning. The underpinning of the technology shifted in the early 1990s, when the optical amplifier began to mature, and simultaneously the distributed feedback (DFB) laser structures required to create the monochromatic output needed in WDM matured. New designs for filters to separate the closely packed wavelengths also emerged, most notably fiber Bragg gratings, but also a variety of approaches based on planar waveguide approaches.

To: Gary Korn who wrote (1290)	2/20/1998 2:42:00 PM
From: Maverick	Read Replies (1) \| Respond to of 12623

WDM: NA, part V As these technologies advanced, the number of wavelengths that could be supported within the 1550 nm transmission band increased. In 1994 and 1995, Pirelli and Nortel supplied four-channel systems, and IBM introduced a 20-wavelength system; 1996 saw the introduction of Ciena's 16-channel system, with other firms subsequently rising to the same channel count. More recently, 32- and 40-channel systems have been announced by several firms for 1998 availability. Advances in WDM are now combined with the advancing pace of time-division multiplexing (TDM). The capacity of each WDM channel is rising, slowly, from 2.5 Gbs to 10 Gbs. The result is that the realizable transmission capacity of a single fiber will have soared from 5.0 Gbs in early 1995 (two wavelengths each carrying 2.5 Gbs) to 100 Gbs or more in 1998. This is shown in Fig. 1, which plots the maximum realizable transmission capacity of single-mode fiber using commercially available WDM and TDM telecommunications systems. Why Use WDM? It is not sufficient that the new technology offer a vision of future capacity; it must offer short-term economic advantages. In practice, there are three basic alternatives that carriers can consider in adding capacity to their networks: adding more fiber; increasing the baud rate of transmission; and now, dense WDM in combination with TDM. The selection of which mechanism (or combination) to use will be complex, but always based on economic considerations. Because fiber deployment is both costly and immensely time-consuming, carriers typically want to do this just once. In practice, all long-distance networks have added considerable amounts of fiber to their original backbones, but the fact remains that this alternative is one that lacks the appeal of using TDM or WDM to expand the capacity of an installed network of fibers. The "traditional" path to upgrading the capacity of transmission networks has been to increase the baud rate of transmission. Over the past 15 years or so, bit rates in use in long distance networks have been increasing at a relatively steady pace: a fourfold rise in capacity every four years. Transmission systems operating at 2.5 Gbs (OC-48 in SONET parlance) have been on the market since 1991 and have become the basic building block of the networks, implying a need for their successors in 1995. There were, however, serious problems in moving to 10 Gbs (OC-192) as the next stage of network evolution:

To: Gary Korn who wrote (1290)	2/20/1998 2:44:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part VI The engineering problems involved in the systems delayed their emergence (although Nortel shipped these systems successfully in 1997, and they are now in use at Worldcom, MCI, Qwest, and other carriers; Alcatel, Hitachi, and Fujitsu are also now on track for OC-192 shipments). Some amount of the installed base of fibers in long distance networks may be unsuitable for use at 10 Gbs. This is not really surprising, since they were not characterized at these new bit rates. Chromatic dispersion continues to vex high-speed systems, but can be overcome with dispersion-compensating devices (The effect results from the fiber having a slightly different refractive index for each wavelength; so a pulse, which must have constituents from a range of wavelengths, is broadened as it passes through the fiber.) One effect that can limit achievable spans of 10 Gb/s systems is polarization mode dispersion (an effect, linked to the residual ellipticity of the nominally circular fiber core, that provides different effective path lengths for the orthogonal polarizations, with the result that the two polarizations slowly separate, making signal recognition more difficult). Another limitation arises less from the fibers than from reflections at cleaved fiber surfaces in the optical path. Some older splices involved flat cleaves, which are detrimental because they generate reflections that in turn can disturb the laser's output. Span engineering becomes more complex as bit rates rise, since receiver sensitivities are typically lower at higher bit rates. This implies that signals need amplification or regeneration sooner. This is a source of considerable inconvenience and cost in building a new network; a carrier that has already placed its amplifier/regenerator huts along its routes is likely to regard their spacing as part of its engineering rules. Furthermore, there is skepticism that a next generation of TDM systems (another fourfold leap would take us to 40 Gbs, OC-768) can be created in a timely and economic fashion. In contrast, the expansion of WDM capabilities to 40 wavelengths and beyond creates almost open-ended scalability of network capacity. While this article's brevity forbids any thorough comparison of the economics of WDM and TDM deployment options, one aspect of the comparison bears repeating.

To: Gary Korn who wrote (1290)	2/20/1998 2:47:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part VIII Furthermore, we have described the use of WDM as a tool to relieve congestion in long distance networks. In practice, it does not take a massive multiyear effort to bring the level of fiber utilization in long distance networks down to the levels of a year or two ago. Therefore, two key questions that must be answered to understand this market's near future direction are: What will prevent this market from being a "bubble," a headache pill that takes care of the headache, and then the problem goes away for several years? Will WDM make sense in short-haul applications? The Future of the WDM Market Expenditures on WDM have soared, from nothing to more than $1 billion in three years. What suggests that it will continue? What factors will prevent WDM from solving the fiber exhaust problem and then retreating? Our firm's analyses suggest three factors: The emergence of short-haul uses of WDM The emergence of other drivers for use of WDM (while the original driver remains intact) The migration toward full network functionality First, Short-Haul WDM Vendors and carriers have been engaged in cost-modeling exercises comparing WDM to traditional SONET TDM upgrade paths for short-haul and mid-length mesh networks. Bellcore analyses presented at the National Fiber Optic Engineering Conference (NFOEC) '96 in Denver, Colorado, suggested that using WDM in the interoffice trunking mesh could save 30 percent over the same network built using SONET systems. In fact, subsequent analyses from participants in this country's major tests of optical network technologies (NTON, National Transparent Optical Network and MONET, the Multiwavelength Optical NETwork) and studies from our firm support this conclusion: short-haul WDM can be cost-effective. We have concluded that this can be the case at bit rates as low as OC-3 per wavelength. This means that carriers will have a whole new way to serve high-capacity clients with seemingly open-ended capacity. And, consistent with this vision, Lucent, Ericsson, Ciena, Cambrian, Pirelli and Alcatel, have all announced their views of metropolitan WDM systems. All have taken steps (e.g., lowering use of EDFAs) to reduce the costs of the WDM systems to equip them for the more cost-sensitive demands of short-haul applications. Several of these systems are passing through (or have already gone through) Bellcore's OSMINE process, and should be ready for use in the Bell system in 1998. We expect that short-haul WDM will be a commercial success, but do not anticipate that its revenues will rise quite as rapidly as did those for long-distance WDM systems.

To: Gary Korn who wrote (1290)	2/20/1998 2:50:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part IX The Emergence of New Drivers for WDM o, we pass to the question of what happens now to long-distance WDM? The prospect of turning each physical fiber in a carrier's network into 16, 32, 40, or more virtual fibers makes a mockery of the fiber utilization or exhaust statistics presented above. A carrier with a network that is 80 percent lit can consider that, given the advent of 40-wavelength WDM systems, its network is effectively just 2 percent lit! A small carrier that deploys WDM faster than its larger peers can find itself managing a network with more capacity than theirs. To that extent, WDM goes far beyond relieving the current level of fiber congestion. Instead, it reverses the current dynamic in the market: backbone bandwidth supplies will be boosted far beyond current demand. Instead of telling their best customers that backbone bandwidth is in short supply, carriers will be working with the customers to entice traffic into the networks. In effect, the deployment of WDM now creates the basis of a tidal wave of available bandwidth coming to the long distance networks, and the result should be a surge of new services and new traffic. This contributes to the idea of WDM as more than a fix to a problem, and positions the new technology as the underpinning of the accelerated emergence of high-bandwidth services. Other benefits of WDM systems have more direct impacts on markets. AT&T's adoption of WDM seemingly is directly tied to the carrier's decision to migrate to a multivendor environment. For this carrier, it becomes possible to use the transponder (in this case transponders e from Ciena and Lucent) to segregate layers of the network, enabling equipment from the dominant SONET terminal vendor (Lucent) to coexist with that of newcomers (AT&T now is looking to use the SONET products of NEC as well as Lucent's). And the coup de grace for WDM is yet to come. For the emergence of WDM into a full logical network layer (either a sublayer below layer 1, SONET, or a competitive alternative to that technology within layer 1) means that it will become increasingly possible to launch traffic from higher layers (2, ATM; Frame Relay; or 3, Internet Protocol, IP) into a self-healing optical network, even without the intervention of traditional SONET network elements. This underpins the approach of Cambrian, which has been exhibiting its products in direct combination with a Newbridge ATM switch: look, no SONET! And router startup Avici describes its future world as not needing the intervention of SONET to launch its products into the public network.

To: Gary Korn who wrote (1290)	2/20/1998 2:54:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part X A development related to this is the subtlety of new-generation WDM systems, which can read (and may, in the future, write) bytes in the SONET header, thus preserving some key benefits of SONET technology, mandating that SONET framing be used, but not actually requiring the presence of SONET network elements. Of course, TDM (SONET in North America, synchronous digital hierarchy -- SDH -- elsewhere) will remain a huge business, and an important part of public network technology for years to come. But the emergence of a logical optical layer over the next several years means that optical network products will be increasingly cost-effective in an increasing range of applications. For this to happen, WDM products must mature to support real optical networks -- which truly is not what current-generation systems execute. From WDM to Optical Networks WDM currently involves almost exclusively point-to-point systems. However, these by themselves are not optical networks but a convenient multiplexing scheme, a tactical means to combine signals from different network elements (such as SONET terminals), leaving the real heavy lifting of network functionality to SONET while performing an interesting optical-layer trick. This, surely, is about to end. Static add-drop systems have been announced by several vendors. These products enable a small number of contiguous wavelengths to be added and dropped at a single site without demultiplexing the entire 16-, 32-, or 40-wavelength bundle. These products are expected to provide increasing value as WDM migrates from long-distance applications to support of metropolitan networks supporting high-speed drops to key customers or to central offices within the city. Later generations of add-drops will provide more dynamic capabilities: dynamic selection of which wavelength(s) are to be added or dropped at a site, or even a full dynamic matrix switch approach. Meanwhile, optical switches are emerging from laboratories into field trials. The systems are demanded as a means to provide cost-effective restoration to the optical network, and will also enable the creation of truly managed networks in the optical domain. The MONET trial (run largely by Lucent, Bellcore, and AT&T) was just one sign that optical cross-connects are emerging. Lucent's decision to bring a product based on that trial's work to market has been followed by others: Optivision has been involved in trials with Sprint, using optical switch technologies to provide optical-layer restoration. Hitachi has performed a field trial with MCI, deploying five optical switches in MCI's network in the Dallas, Texas area. These switches acted as a logical self-healing mesh network. Optical switches themselves will experience rapid evolution, one of expanding complexity as their roles move from bulk restoration of multichannel optical network paths to becoming the key agents of wavelength management in the evolving optical network.

To: Gary Korn who wrote (1290)	2/20/1998 2:55:00 PM
From: Maverick	Respond to of 12623

WDM: NA, part XI The Future s divestiture broke up the monopoly of the Bell System nearly 15 years ago, fiber systems were working at the then unheralded rate of 90 Mbs. New systems boasting as much as 410 Mb/s emerged soon; systems soon cracked the gigabit barrier, and advanced to 2.5 Gb/s, and now to WDM, with single-fiber transmission capacity already at 80 Gbs. Each advance had some wondering how much capacity is enough. We now know that the computers on our desktops and the public and private networks to which we attach them are spiraling around each other, one demanding more bandwidth, the other supplying it. This dance will go on. Earlier in this brief review, I mentioned in passing the lack of confidence that 40 Gb/s (effectively OC-768) would come to pass in the foreseeable future over existing networks. Certainly, there are grounds to think that the engineering tasks associated with this are truly daunting. However, vendors are preparing generations of systems combining dense WDM with 10 Gb/s technologies to bring about another great step forward in fiber throughput; Nortel is already delivering systems that combine eight wavelengths with 10 Gb/s per channel. At NFOEC '97 in San Diego, California, Fujitsu, Nortel, and Hitachi described their plans to bring out products with 16 or 32 wavelengths each at 10 Gb/s, an amazing 160 or 320 Gb/s per fiber! Who would ever need more? In fact, there are already moves afoot to do just that: to expand the capabilities of fiber and optical networks. These efforts explore new combinations of technologies, including WDM and TDM techniques, new modulation schemes, new schemes to mitigate the performance-sapping effects that occur when fibers are no longer purely transparent, and perhaps even new generations of optical fibers. The objectives of these endeavors is to establish those combinations that will enable us to achieve terabits per second per fiber early in the new decade. Specifically, MCI has declared that it wants to transmit 32 wavelengths, each carrying 40 Gb/s, on its fiber backbone: 1.2 Tb/s on one fiber!

To: Gary Korn who wrote (1290)	2/20/1998 2:57:00 PM
From: Maverick	Read Replies (2) \| Respond to of 12623

WDM: NA, part XII "Open" WDM Systems Commercial WDM systems exist today in three basic point-to-point configurations: With the wavelength-specific transmitters (and receivers) included in the SONET systems (or other network elements) With the wavelength-specific transmitters incorporated into a transponder (that receives signals from the SONET or other systems), but with the receivers at SONET systems With both wavelength-specific transmitters and receivers incorporated into transponder banks The transmitter configuration described in Fig. 2 is of the latter, "open system" configuration. As described in the text, SONET and ATM systems feed the transponder terminals and, at the far end, a receiver bank separately detects each wavelength's signal and a short-haul laser is used to retransmit the signal to the final receiving network elements. The objective of this approach is to enable the optical layer to support transmission of signals from dissimilar network elements, including systems that are not typically configured for long distance telecommunications.