Lightera Units Replace SONET AD Muxs Via Smart Signaling & Routing Bandwidth Management - Omnia Provides Local Optical Rings to Enable Optical Transport Through the Local Loop and Across the WAN,

pp. 1 - 10

We interview Joe Berthold Vice President of Network Architecture and Standards. Starting in 1996 Ciena introduced wave division multiplexing and then dense wave division multiplexing (DWDM). The early part of the interview covers technical issues such as how Bragg Filters are made and used in DWDM as well as the characteristics of 1300 versus 1500 nano-meter fiber.

In 1996-97 Ciena worked with Sprint to help alleviate that carrier's fiber shortage. They developed their optical amplifiers and Bragg grating filters in house. The carrying capacity of fiber is being refined every year in order to better match the changing characteristics of the amplifiers and switching devices available. This is the reason that Qwest, Level 3, and Williams only fill one of ten or more conduits in the right of way at a time. The premise is that the time the first conduit is used up, there will be much better fiber to lay in the second conduit.

For integrated network management they have used a device called the Sentry 4000 which was introduced in January '98. In this three shelf equipment rack you have you network management hardware and an optical amplifier.. The second shelf supports eight different transmitters. The final shelf contains receivers, each with a Bragg grating matched to a different color. To add channels, they add cards. Depending on what wavelength they want to turn up, they put in a transmitter with the desired frequency, and a receiver with the appropriate Bragg filter. While they offer both OC-48 and OC-192 interfaces now, OC768 doesn't appear practical commercially because, with current fiber, they would have to reamplify it so often that it would not be economic to do so. Moving now from OC 48 to OC 192 looks practical because the electronics on the ends of the fiber does not cost four times as much.

One of Ciena's developments has been an all optical network that removed the need for electrical regeneration of the laser pulses and therefore the need for SONET add drop muxs that cost any where from several tens of thousands of dollars to several hundreds of thousands of dollars apiece.

Lightera, which Ciena purchased early this year, will enable Ciena to bring to market by year's end a device called the Core Director. In one equipment bay it will be possible to terminate 256 different OC-48 (2.5 gigabit) channels using OC-48 plug-ins each the size of two business cards. If a Core director were configured to behave like SONET Add Drop Multiplexers, one unit could replace 48 bays of SONET equipment. While the Core Director can emulate many Add Drop multiplexers, it can also, at the same time, operate as an optical channel switch or optical crossconnect.

The Core director is configured on a port-by-port basis. Is this port part of a SONET ring or not? Or is this port just a port on a switch that is part of a mesh network? Do you want to use restoration? Ciena has a set of restoration protocols built into this, called optical signaling and routing protocols that allow the Core Director to behave in a similar way to a network of ATM switches, automatically reconfiguring in the event of loss of resources somewhere.

They have developed protocols that will allow the core Direct to perform both the functions of layer 2 switching and layer 3 routing. Their OSRP (Optical Signaling and Routing protocols are designed to make bandwidth provisioning easier. A circuit can be activated by installing a small plug in place a point and click on a GUI interface using their network management software.

Corestream introduced in June 99 replaces the Sentry 4000 and comes with an initial configuration of 96 channels at 2.5 Gigabits or 48 channels at 10 G, or a mixture of both. The architecture is extensible to a speed of two terabits per second. A further stage of evolution called "Multiwave Lightworks" will integrate the Core Stream technology with the Core Director, so that the WDM interfaces are inside the same platform, saving both additional interfaces between the two, and space, power, and cost.

Ciena is also moving into provisioning equipment to bring high speed optical streams through the local loop to customer premises equipment. Smaller versions of its wide area network products will allow carriers to terminate new broadband circuits without having to set up new SONET rings or buy $100,000 SONET AD multiplexers for individual buildings.

Omnia is another company acquired by Ciena earlier this year. The Omnia AXR 500, is a service delivery and multiplexing box which makes it very simple for a carrier to offer a number of services ranging all the way from voice up to ATM to IP to private Ethernet networks. The AXR500 provides service interfaces and then multiplexes the services and bring them back to the carrier's office.

Ciena may be the only player which currently has a combination of long-haul transport systems, point-to-point systems in Metropolitan Areas, and finally ring access systems. Its strategy is to be able to originate a wavelength at a customer location, route it to an office, hop onto a long-haul system, go across the country, come back out to its local system, and go back out to the customer. It wants that entire transport from customer to customer to stay as an optical channel.

Sean Donelan on Breaking into National Cost Free Peered Status,

p. 10, 20

Sean Donelan points out that it has been several years since a large national or international ISP has been able to become a top level national peering player without buying another ISP that was already fully peered.

Role of Satellite Bandwidth in Delivery of Content Across the Global Internet

PanAmSat Discussion Explores Economics of Content Delivery, Decreases in Cost of Receiving Equipment Make Delivery of Same Data to Many Different Places Competitive with Fiber

pp. 11 - 18

We interview Rob Bednarek, CTO, Gail Fell, New Media Director and Aaron Falk Protocol Specialist. The interview starts with and overview of the satellite industry focusing on the declining cost of receiving equipment and the different market position of GEO's and LEOs.

Satellite's major niche in the Internet business model is the delivery of content to multiple sites. Such content may be continually updated web caches at major ISP backbone sites or delivery specialized content to a receiving business with many locations scattered over an entire continent. While fiber can reach up to about ten different sites economically, if the number of sites which must get the same material increases beyond ten and into the hundreds the advantage swings to satellite's favor.

Note also that for connecting ISP's in remote areas in a single hop to a global backbone, they are quite good. Any content for something like an advertising campaign, or something like a Victoria's secret fashion show that has to be delivered in mix and match proportions to servers all over the world is a good candidate for satellite delivery.

In the area of protocol development there are several proprietary solutions to reliable multicast, that are currently available. They are developed by independent companies which have some software-only and some software-hardware combined solutions for doing reliable multicast transmission, targeted mostly toward satellite. You have a single sender and a bunch of receivers and it's one hop. And there are a variety of techniques that are being used for forward error correction to allow scalability to fairly large numbers of receivers.

According to Fell reliable multicast is a way of taking information and sending it one to many multiple destinations in a way that ensures error-free delivery. There's a variety of ways that one can ensure error-free delivery. In Unicast, TCP is used. TCP acknowledges packets actually received. The problem is that if one has a lot of receivers and they're all sending acknowledgements to the sender, then the sender can easily get overwhelmed, so that approach, the acknowledgement-based approach, does not scale well to large numbers of receivers. But there is a variety of ways that the problem may be solved. For example one may inhibit a acknowledgement transmission and then send the file many times. Doing this will increase the certainty of reception by all sites. One of the strengths of the satellite industry is that because of its topology, it can get to everyone. A user doesn't need to worry about different tariffs or about signals using another provider's resources, or about crossing provider boundaries for its customers?

The Internet protocol has helped the standardization of satellite data protocols far more than anything else. Because from the physical layer, it is possible to adapt the necessary things on satellite receivers, but the choke point has been what happened after that and, traditionally, physical satellite networks were used in very closed architecture data networks. Such networks tended to be private networks, with the exception of voice, they were not part of big public networks. However, IP changed all that. As a result, the request that comes from a foreign country is, I need a pipe for IP interchange and I'll take care of everything downstream from the physical layer in my country, you just get that connectivity to me.

For third world ISPs satellite delivery from a cost standpoint, it's fairly constant, and, actually, that's one of the appeals, is it's entirely within the control of the ISP. An ISP can contract for as much or as little connectivity to the U.S. as is required to satisfy its customers. Different ISP's have different opinions as to how long customers should wait for response times. This is something that lets them differentiate themselves from each other.

If you can go from having ten people receive your bits to having a thousand or a million people receive them,, your actual price per bit is going to change considerably. The key becomes defining a set of services that can take advantage of the natural multicasting which satellites provide.