SI
SI
discoversearch

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 : MRV Communications (MRVC) opinions? -- Ignore unavailable to you. Want to Upgrade?


To: Regis McConnell who wrote (13959)6/18/1999 9:55:00 AM
From: Greg h2o  Respond to of 42804
 
I would imagine this step toward looking a little less independent is just one in a line of steps.... we'll see....



To: Regis McConnell who wrote (13959)6/18/1999 10:37:00 AM
From: Sector Investor  Read Replies (1) | Respond to of 42804
 
<<600 to 800 million, eh? Sounds like a good place to start for Charlotte's Web.>>

I am back - refreshed after a killer 24 hour (plus 11 hour time zone difference) commute.

Yes, my first thought is that this does a few things:

1) Verifies that LU sees the need for Terabit products
2) Gives an idea what LU thinks Nexabit is worth and sets a general value range for Terabit companies
3) Shows what a steal MRVC is getting, as Charlotte's Web's products are strong ones too.

Before I left, I started a spreadsheet on Terabit Routers and passed it on to Ron Stange for update/expansion while I was gone. He apparently shipped it to Jack, and now I have an updated version back (neat huh?)

There is a Nexabit column now, as well as several others, but there are still many blanks (more work for the thread to do). I haven't looked it over closely yet, but I will tonight or tomorrow and post the MRVC-Nexabit comparison as it now stands.

Everyone is welcome to help us fill in the blanks.



To: Regis McConnell who wrote (13959)6/18/1999 2:44:00 PM
From: Ronald D. Stange  Read Replies (1) | Respond to of 42804
 
Regis:

These are looking a bit less independent, no?. Yes

First saw the Charlotte's link a week ago and now there is the NAC link. My talks with Mannix O'Conner at I+P about increasing share holder value was associated with the possibility of IPOs for NAC/CW if that arrangement would provide the vehicle for accomplishing that objective but any decision would depend on demand for NAC/CW products and/or partnering of those affiliates with other entities.

Ron




To: Regis McConnell who wrote (13959)6/19/1999 1:05:00 PM
From: Sector Investor  Respond to of 42804
 
On Wednesday, I see Frank Coluccio posted a very interesting and important article on short haul DWDM and the thread totally ignored it. No mention of MRVC or New Access of course - typical, but New Access is a very new player.

Sloppy, sloppy. Sector goes away for 2 weeks and the thread develops bad habits. Well I'm back and it's time to crack the whip.

Which brings up the old joke:

[Female voice] "The whip! The whip! Anything but the whip!

[Male voice] "ANYTHING??????"

[Female voice] "The whip! The whip!"

We have been concentrating too much on the Terabit routers and neglecting New Access and Metro DWDM prospects. I mean you DO want to know what MRVC's New Access's prospects are don't you? It's obvious the market doesn't know just yet, but we need to know.

I'm sure that even non technical people (if you take the time to read this), should be able to see the economics, the dynamics of this market and the HUGE OPPORTUNITY out there for superior solutions.

I am going to repost this very lengthy article, but I'm taking the time to break it up into several posts and highlight parts via bolding and Upper Casing (so everyone will actually read it). I also added a few comments enclosed in brackets [ ].

This article raises many interesting points, and should provide for significant discussion on all threads (well two of them anyway).

The key things to keep in mind as you read all this (please do) is "What does this mean for New Access's prospects? What pieces of information do we need to know?

Comments not just appreciated, they are requested.

OK here goes



To: Regis McConnell who wrote (13959)6/19/1999 1:07:00 PM
From: Sector Investor  Respond to of 42804
 
DWDM: In for the Short Haul?

The promise of just-in-time provisioning makes now the time to assess metro DWDMs

Carriers are sold on DWDM's ability to go long. It's the short game they're not so sure about. While the technology has proven its punch in delivering big amounts of bandwidth over long distances, it's proving a poor fit for big cities.

PRICE IS THE PROBLEM. "It's cheaper to pull new fiber in metropolitan areas than to light up a DWDM wavelength," says Tim Weingarten, an equity research analyst at BancBoston Robertson Stephens Inc. (New York). And not just a little cheaper, either: Some carriers say the price of DWDM (dense wavelength-division multiplexer) equipment will have to come down at least 40 percent before the technology is worth deploying in cities. [Sector note: New Access addresses this by using only one wavelength per node - major cost savings - much more than 40%!]

But carriers that can't afford it now can't afford to ignore it either. That's because DWDM could be the key to the grail of carrier services: just-in-time provisioning of bandwidth.

Several vendors [MRVC included] are working on so-called optical switches that route wavelengths from one fiber to another; when combined with metro DWDMs, they could ultimately let carriers assign an OC48 (2.488-Gbit/s) circuit to a customer in SECONDS-rather than the MONTHS it now takes to build out new Sonet rings.

So what's the best way to prepare for metro DWDM? Follow a two-pronged plan. One part involves scouting out the gear: DWDM devices differ on everything from the number of wavelengths and the distances handled to interfaces and provisioning capabilities. And though functionality is low and prices are high, that's bound to change as vendors fine-tune their wares. [Sector note: this is happening now]

The other part is to build the business case. Fact is, metro DWDMs could actually be an economic alternative to laying new fiber in large, densely populated cities like New York and Los Angeles. "If you're in a congested metropolitan area with very crowded infrastructure, then DWDM would be the cheapest way to enhance capacity," says Mark Chall, director of network evolution planning at Sprint Corp. (Kansas City, Mo.). Ask Peter Tierney, chief operating officer at CLEC (competitive local exchange carrier) Millenium Optical Networks Inc. (New York). Rather than building multiple Sonet rings-which New York labor costs and construction regs would make way too expensive an undertaking-he plans to deploy an eight-wavelength system from Sycamore Networks Inc. (Chelmsford, Mass.).

And carriers seem to be realizing the potential payoffs. "We think the market is really going to take off next year," says Dana Cooperson, senior analyst at Ryan Hankin Kent (RHK, South San Francisco). All the more reason to sweat the metro DWDM details now-before the competition does.



To: Regis McConnell who wrote (13959)6/19/1999 1:08:00 PM
From: Sector Investor  Respond to of 42804
 
Riding On the Metro

Right now, there are five [make that 6] vendors poised to take advantage of that market: Ciena Corp. (Linthicum, Md.), Ericsson AB (Stockholm, Sweden), Northern Telecom Ltd. (Nortel, Mississauga, Ontario), Sycamore, and Osicom Technologies Inc. (Santa Monica, Calif.) (see Table 1). Long-haul DWDM vendors Alcatel S.A. (Paris) and Lucent Technologies Inc. (Murray Hill, N.J.) plan to ship metro devices later this year. And startup Optical Networks Inc. (ONI, San Jose, Calif.) says it has a DWDM product set for beta shipment in August.

But what exactly is it that can make metro DWDM prohibitively expensive? The answer lies in the technology itself.

DWDM was invented to boost bandwidth on long-haul fiber optic networks. Instead of carrying data through the fiber cable on a singlewavelength-or color-of light, DWDM sends multiple colors through the same cable. Each wavelength typically carries 2.5 Gbit/s. For long distances, the technology been very successful: Sprint now uses DWDM in about 80 percent of the fiber in its long-distance network; AT&T (Basking Ridge, N.J.) and MCI Worldcom Inc. (Jackson, Miss.) post similar numbers.

But DWDM in metropolitan areas makes for a different ball game. For one thing, metro fiber systems are organized in rings, rather than point-to-point circuits. That means traffic must be added to the ring and dropped off at multiple nodes around the circumference, requiring a reengineering of equipment to accommodate that capability. [Metro Fusion has this capability]

For another thing, DWDM does not eliminate the need for amplifiers in short-haul fiber optic trunks the way it does over the long haul. Most metro area links run only 120 km at the most-as opposed to the 1,000 km that long-haul circuits do-so no amplifiers are needed anyway (or if they are, they're relatively cheap). In other words, the DWDM incentive disappears; pulling new fiber in metropolitan areas is actually less expensive than installing a whole new system. [ but this may not be true with MetroFusion's lower cost structure]



To: Regis McConnell who wrote (13959)6/19/1999 1:10:00 PM
From: Sector Investor  Respond to of 42804
 
Different Wavelength

Expensive or not, metro DWDM is something carriers have to come to grips with. And assessing metro DWDMs knowledgeably means knowing how they resemble Sonet gear-and how they differ.

The devices are similar to Sonet (synchronous optical network) devices in that they sit at various nodes on a fiber ring and drop traffic into it to be transported to the other nodes. But they're different in that they divide the ring's capacity into multiple wavelengths, rather than splitting traffic into time slots on a single wavelength. And that makes for a massive boost in speeds: While a Sonet ring can theoretically scale to a maximum of 10 Gbit/s, the most powerful metro DWDM product can scale to 110 Gbit/s. (The most advanced wavelength multiplexing system right now is Lucent's Wavestar OLS 400G, which is designed for long-haul networks and can transport up to 80 colors of light on a single fiber strand for 400 Gbit/s of total throughput.)

The technologies also differ in terms of deployment. Scaling Sonet above maximum capacity means building multiple rings in metropolitan areas (see the figure). Carriers interconnect their central offices (COs) via several OC48 rings in a configuration known as an interoffice ring; a typical incumbent local exchange carrier (ILEC) in an average-sized city, for example, may hook up four COs in this way. From each CO, they then build out access rings to corporate customers and other carriers, such as ISPs (Internet service providers) and long-distance operators. The access rings, which today run at OC3 (155-Mbit/s) or OC12 (622-Mbit/s) speeds, typically support four nodes each.

Further, each ring is built with two fibers, so that traffic can be rerouted from one to the other in the event of a fiber cut. Fiber cables from the various rings usually travel in the same maze of underground conduits; a conduit carries up to 400 fiber cables, according to Banc Boston's Weingarten.

Carriers terminate circuits at the customer premises by pulling fiber into the basement of the building, where they install Sonet ADMs (add-drop multiplexers). The ADM furnishes DS-1 (1.544-Mbit/s), DS-3 (45-Mbit/s), and OC3 services to the customers; traffic from those lower-speed interfaces is packetized into Sonet frames and "dropped" into time-slots on the ring; they are pulled off at the destination node by another ADM.

The trouble with Sonet right now is that most rings are more than half-filled with traffic. Every time a customer orders another OC12 or higher, a carrier has to build out another ring from its CO-a process that takes a minimum of six months. Even merely upgrading the ring to a higher speed could take months.
[The DWDM advantage here should be obvious]



To: Regis McConnell who wrote (13959)6/19/1999 1:12:00 PM
From: Sector Investor  Respond to of 42804
 
Riding The Wavelengths

And that's where DWDM enters the picture. Instead of pulling new fiber, vendors say, why not upgrade the whole ring with DWDMs? They sit in the basement of the customer premises, just like Sonet ADMs. Two fibers are typically pulled into the DWDM chassis from the metropolitan ring; as with Sonet, one is used as a so-called primary circuit, and the other as the "protect," or backup, circuit (see Figure 2).

The DWDM chassis has multiple slots for cards. Each card allows the unit to drop one or more wavelengths onto the ring; it also contains single-mode and multimode fiber interfaces for delivering services to end-users.

And the number of wavelengths is what helps separate current products. Sycamore's SN8000, for instance,
handles the most: 44. Ericsson's Erion Networker Flexing Bus handles the fewest: just 16.

Distance is another differentiator, and in the metro area it's measured in terms of the ring circumference, or the total distance from the CO to the furthest node and back again. Ericsson and Sycamore claim to support
rings as large as 500 km, thanks to optical amplifiers. Ciena's Multiwave Metro doesn't use amplifiers, and the maximum distance it handles is 80 km. Does that mean amplifiers are the way to go? Not necessarily, says James Rouse, director of market development for optical products at Nortel. His research indicates that 90 percent of rings in the U.S. are under 120 km in circumference, which RHK's Cooperson confirms. Because amplifiers cost a lot of money, both Nortel and Osicom offer them as slot-in options. They can be removed when deployed on shorter rings, keeping down costs. Nortel says its Optera Metro can go to 120 km, and Osicom says its Gigamux reaches 300 km with amplifiers.

Plug and Pay Less

As for provisioning services, carriers can tackle the task two ways. They can bring circuits to customers either directly on top of wavelengths, or they can carry them using current Sonet gear.

In the first scenario, customers just plug their equipment into the DWDM, and each of the services they use travels natively over a 2.5-Gbit/s wavelength, with multiple wavelengths used to provision multiple services. In the second scenario, carriers connect a DWDM to a Sonet ADM. Customers plug their equipment into the ADM instead, using its multiplexing capabilities to break down the 2.5-Gbit/s wavelength into several lower-speed services, like T3s (45 Mbit/s) and OC3s.

Whether one approach is better depends on the service being sold. Splitting wavelengths means "you don't have to waste a whole wavelength for a relatively low-speed signal," explains Ron Mackey, Osicom's executive vice president of technology. But big-bandwidth apps work at such high speeds there's no need to multiplex them down, so they're better handled without additional Sonet equipment, says Steve Chaddick, Ciena's senior vice president of strategy and corporate development. He thinks any service above OC12 is delivered more efficiently on top of wavelengths. Then again, carriers might want to keep their Sonet add-drop multiplexers and use them in conjunction with a DWDM. "Sonet is something CLECs and RBOCs have lived and breathed for a long time, and they love it," Weingarten says. Most of today's DWDM products, he adds, don't offer nearly as much as Sonet in terms of configuration, restoration, or management.

Osicom, however, says its Gigamux may offer the best of both methods. It has so-called EPC (electrical
photonic concentration) cards that split each wavelength into 16 lower-speed pipes supporting such services as DS-3, OC3, fast Ethernet, and FDDI. Baksheesh Ghuman, senior product manager at Electric Lightwave Inc. (Vancouver, Wash.), a CLEC that has been testing the product, says that's made for a 20 percent savings over a Sonet ADM. What's more, the Gigamux bundles everything in a single device, making it easier to manage. "Service providers like solutions," says Deb Mielke, an independent consultant at Treillage Network Strategies Inc. (McKinney, Texas). "They don't like to have to put everything together all the time." But Osicom's multiplexing scheme is different from Sonet's, so carriers may need to retrain their engineers in order to do that management.




To: Regis McConnell who wrote (13959)6/19/1999 1:13:00 PM
From: Sector Investor  Respond to of 42804
 
Pros and Cons

But even if DWDMs are used without Sonet, they can offer a lot of different services. All products deliver either an OC3, OC12, or OC48 pipe over a single wavelength. Ciena, Ericsson, and Nortel also offer a native gigabit Ethernet port, allowing carriers to sell heavy-duty transparent LAN services to big-spending corporations. And Ericsson, Nortel, and Osicom also supply lower-speed services like fast Ethernet and FDDI, which run at 100 Mbit/s. [I believe Metro Fusion does too]

Carriers also should consider whether a DWDM allows them to run such high-margin apps as Escon (Enterprise System Connection), Ficon (Fiber Connectivity Channel), and Fibre Channel. Escon and Ficon are mainframe protocols used by banks, insurance companies, and large utilities; they run at 200 Mbit/s and 1.06 Gbit/s. Fibre Channel, which is used to build storage-area networks, runs at 1.06 Gbit/s.

Right now, only Nortel's Optera actually allows provisioning of Escon, Ficon, and FibreChannel. While Osicom's Gigamux can handle them, it can't manage or bill for them-which means no provisioning. [I don't know about MetroFusion - we need to find out]

Carriers might be wondering why there's a need in metro areas to carry data at such high speeds in the first place. "Financial institutions cannot afford to keep all their customer data transactions in one location," explains Brian McCann, president of ADVA Optical Networking Inc. (Ramsey, N.J.), which makes point-to-point DWDM systems for such purposes. "It must be fully backed up in real time." Other important apps include video transmission for media companies, typically handled by the D-1 transmission scheme at 270 Mbit/s, or by HDTV (high-definition television), which runs at 1.5 Gbit/s.

In fact, DWDM's ability to carry such enterprise applications makes for one more compelling advantage over Sonet. Traditional Sonet gear can't easily carry Escon, Ficon, and Fibre Channel because it operates at fixed rates (OC3, OC12, and OC48); Fibre Channel, remember, operates at 1.06 Mbit/s.

And selling a variety of services via metro DWDMs is one way carriers could really make a killing. "I want to provide one customer with different services running on different wavelengths," says Millennium Optical's Tierney. "On channel 1 they'll get gigabit Ethernet, on 2 is an ATM service, on 3 a dial service, and on 4 an Escon application." This kind of bundling not only brings in more revenue, he adds, but also it helps keep customers and reduce churn. [We need to find out how MetroFusion addresses this]

But can carriers beat corporations to the punch? Many businesses are leasing dark fiber and buying point-to-point DWDMs to furnish such services for themselves. "Carriers cannot sit back and just watch them do this," warns McCann. "They need to sell services, not pipes." Some carriers are selling dark fiber for as little as $5,000 to $6,000 a month. Selling an enterprise application on top of that fiber might pull in$20,000 instead, he says.



To: Regis McConnell who wrote (13959)6/19/1999 1:16:00 PM
From: Sector Investor  Respond to of 42804
 
Declarations of Independence

And allowing for a mix of protocols isn't as complicated as it sounds. That's because most metro DWDMs are protocol and bit-rate independent (Sycamore's is the exception). Customers plug their equipment into a single fiber interface; the device recognizes the bit rate and service coming in and notifies the management console. The console checks its database to ensure the customer has paid for that amount of bandwidth, and then maps the service onto the appropriate wavelength.

Then again, not all gear lives up to its independent billing. Ciena and Nortel, for example, give customers a choice of two fiber interfaces, one for handling lower-speed services like OC3, OC12, fast Ethernet, and Escon, the other for OC48, gigabit Ethernet, and Fibre Channel. Upgrading between speeds requires cards to be swapped out.

And while Osicom claims to support services of all speeds on the same physical port, the vendor also sells additional modules that work with specific protocols. That's because its universal interface can't recognize which service is being carried, so it's unable to manage and bill for it. Similarly, Ericsson's universal interface handles all services but can only manage and bill for a small number of them.


Give It Up

There's plenty to be gained by switching from one technology to another, but carriers may also wonder what they'll lose by dropping Sonet in favor of DWDM. Fact is, most DWDM vendors are still struggling to replicate the path restoration, management, and time-slot assignment capabilities available in Sonet ADMs.

Start with the ring configuration. In Sonet, rings can be organized according to two Bellcore standards: UPSR (unidirectional path-switched ring), or BLSR (bidirectional line-switched ring). UPSRs enforce a rigid 1:1 redundancy throughout the ring; BLSRs allow some of the redundant capacity inherent in Sonet to be reused.

But most DWDM vendors can't offer BLSR support. That would be a tricky engineering feat, according to Weingarten. In fact, Sycamore is the only vendor to have pulled it off. [We need to ask this of New Access]

As for protection switching, DWDM actually beats Sonet in one way: In the event of a fiber cut, most devices can reroute to the backup fiber in 25 milliseconds or less, whereas Sonet ADMs can take as long as 50 ms. But problems crop up when Sonet and DWDM are used in conjunction. DWDM's own protection switching capabilities have to be turned off to avoid contention between the two fail-over schemes. All vendors let users turn off protection switching-except for Sycamore. It says its fail-over rate is just 3 ms, which means it can switch over to the protect line before Sonet even realizes it.

Generally speaking, Sonet also beats DWDM on time-slot assignment. From a management console, carriers can program a Sonet ADM to assign a time-slot from one customer to another; an internal cross-connect simply switches the circuit over. But with DWDM, there's no way to switch one wavelength assignment from one customer to another without physically moving interface cards between slots in the chassis. The exception is Nortel's Optera, which contains an electrical cross-connect to reprovision wavelengths between customers.

Wavelength assignment is vital because it's one of the requirements for just-in-time provisioning. "Let's say Amazon.com calls its service provider and says it's having a book sale for a week," explains Weingarten. "They might need an extra circuit to handle the increased traffic load." Wavelengths could be reassigned specifically to handle that temporary increased load.

Management capabilities are next on the checklist. Vendors furnish their devices with management software that can talk to various OSSs (operation support systems). Most DWDMs communicate with the OSS through various legacy Bellcore standards like TMN (telecommunications management network) and NMA (network management application). Also, Nortel, Osicom, and Sycamore support Corba (common object request broker architecture) programming tools, which allow DWDMs to be programmed to communicate with OSSs. Generally, incumbent carriers are sticking with the Bellcore specs, while competitive entrants prefer to write Corba code-or even rely on SNMP from the data networking world. "CLECs want to manage everything like one big WAN," says Osicom's Mackey, adding that SNMP lets them do that.



To: Regis McConnell who wrote (13959)6/19/1999 1:18:00 PM
From: Sector Investor  Read Replies (1) | Respond to of 42804
 
All In the Numbers

Still, all the DWDM investigative work could be moot if the economics don't work out. Right now, deployment is worthwhile only in a few major cities-and long carrier buying cycles mean that even in those locations there have been few takers. "I'd say only about six carriers have signed on so far," says RHK's Cooperson.
[Does this sound like an opportunity for an effective low-cost solution or what?]

Look at the cost structure. Metropolitan rings consist of short spans between nodes that average less than 20 km, while long-haul fiber systems run to thousands of kilometers in one stretch. "The key is how much it costs to lay a fiber optic cable per mile," says Sprint's Chall. He estimates that in Las Vegas, his biggest market, it can come to as little as $1,000 to roll out a truck, hire some workers, and pull fiber through an existing conduit. To roll out fiber across one segment of an average metro ring, therefore, costs a little less than $20,000; the carrier then has to tack on additional costs for buying, installing, and configuring a Sonet ADM to service the new customer.

Compare that with DWDM prices. Sycamore's system starts at $20,000 for a base chassis and $48,000 for each new OC48 wavelength. "The numbers don't add up," says Chall. "For segments under 30 km, it's cheaper by and large to roll new fiber than it is to use DWDM." [I wish we knew more about MetroFusion's cost structure]

Fortunately, there are signs that metro DWDM pricing may become more attractive. As with silicon, costs of optical components (like the lasers used to light the wavelengths) keep falling. "The pace at which this technology is changing is just amazing," says Jeff Kiel, Sycamore's director of marketing. "The lasers just keep getting cheaper and cheaper." And don't forget about those cities where the technology probably does make economic sense. "In New York, carriers are allowed to dig only at night," explains Weingarten. Such regulations boost labor costs, and, coupled with the fact that many conduits are already full, makes new fiber rollouts a lot more expensive-up to $20,000 per km, according to analysts. And then there are always the difficult routes. "I've been doing telecom in New York for 15 years and there are places considered the Bosnias of the industry," says Optical's Tierney. "Staten Island, N.Y., is one of them." There are only six fibers linking Manhattan to Staten Island, and that forces carriers to deploy DWDM there.

Finally, net architects should consider the long-term implications of managing wavelengths, rather than time-slots. Although products won't be shipping until later this year, several startup vendors-including Monterey Networks Inc. (Richardson, Texas)-are developing so-called optical switches that move wavelengths on the fly from one fiber ring to another. Instead of taking six months to provision an OC48 circuit, carriers will be able to turn it up in seconds. [We are working on this too]

And it's that just-in-time provisioning of high-bandwidth services that has the industry excited. "Trust me, I'm not one of those missionary-type guys," Tierney says. "But I just got off the phone with the CEO of an Internet service provider who said he'd love to buy bandwidth at off-peak rates. Right now, I have to tell him this is not the airline industry." To make the most of his DWDM investment, he'll sell high-return OC48s to other carriers, allowing them to interconnect their central offices. And as the economics allow, he'll start selling bandwidth to high-rolling corporate customers, possibly in real time. "There's a very profitable business in providing high-bandwidth services in a local area," he says.

Andrew Cray is WAN equiptment editor for Data Communications. He can be reached via e-mail at acray@data.com.