Qwest, Broadwing Go Gigabit Ethernet

LAS VEGAS -- Networld+Interop -- Broadwing Communications Inc. (NYSE: BRW) and Qwest Communications International Corp. (NYSE: Q) today announced plans for high-bandwidth data connections based on gigabit Ethernet networking equipment, giving a nudge to the trend of widely available network services that provide bandwidth as needed.

Until now, high-speed Ethernet services had been the focus of startups such as Yipes Communications Inc., Cogent Communications Inc., and Telseon that are largely serving metropolitan markets. The entrance of major carriers boosts the market's viability -- and lends more evidence that Ethernet is becoming a leading metropolitan data services format (see Metro Optical Ethernet, Ethernet in the 'Hood , and Endless Ethernet?).

Qwest's offering, called Dedicated Internet Access (DIA), is focused on connecting customers at Internet data centers. It will supply Ethernet connections ranging from 2 Mbit/s to 1 Gbit/s. Broadwing is testing a service that connects businesses in metropolitan areas to fiber-optic-based gigabit Ethernet connections on Broadwing's optical backbone, also at speeds up to 1 Gbit/s (see Broadwing Trials Optical Gig-Ethernet).

Although the entrance of the large carriers will put pressure on the younger metro services outfits, which are saddled with high startup costs, in some cases the carriers are showing a need to partner with them. For example, in Broadwing's case, the carrier is expected to announce on Wednesday that it will market its gigabit Ethernet services with the help of Telseon.

The competitive threat to existing metro services is also difficult to quantify because neither Broadwing nor Qwest is disclosing pricing -- but the prices are expected to be significantly higher than the rock-bottom pricing touted by the startups. Each carrier says the gigabit Ethernet services will be priced according to the actual bandwidth that is used by the customer, rather than charging for a fixed-size pipe. Broadwing will measure usage in 1-Mbit increments, and Qwest says it has the ability to measure usage and charge customers by the volume of IP traffic that's used.

Metered usage and the ability to provide guaranteed service standards are considered key elements in making high-bandwith Ethernet services successful, because those features would give carriers a better business case and allow them to charge more.

Officials from both Qwest and Broadwing claim their services will be differentiated from those of smaller providers because they own the backbones and will thus be able to control the quality of service on the network.

"We need to make sure we can get mediation out of the equipment and do usage-based billing," says Tony Tomae, vice president of data/Internet solutions at Broadwing.

"We are selling services to connect to one of the fastest Internet backbones in the world," says Tyler Coleman, product manager for dedicated Internet Access at Qwest. "It may not be the cheapest services, but you will get industry-standard SLAs [service-level agreements] and 100 percent uptime. We're not exactly competing with Yipes."

Industry experts said the development is proof that Ethernet data networking technology, which started in corporate networks, is now becoming reliable and powerful enough for large telecommunications carriers.

"In the next few months we will see carrier-grade service and Ethernet come together," said Mark Allen, co-founder and vice president of telecommunications testing service Valiant Networks.

-- R. Scott Raynovich, Executive Editor, Light Reading
Page 1 / 3   >   >>
metroshark 12/4/2012 | 8:27:43 PM
re: Qwest, Broadwing Go Gigabit Ethernet It would be interesting to see who these carriers choose for their Gigabit Ethernet infrastructure. Will they go with the good old Cisco, or will they choose one (or more) of the Layer 3 switch vendors?
rickwang 12/4/2012 | 8:27:39 PM
re: Qwest, Broadwing Go Gigabit Ethernet in layer 3, the route is more stronger than the layer3 switch. the proformance and relibility,
the switch is only the switch, not the router.
the carrier should make a good aged choice.
gea 12/4/2012 | 8:27:38 PM
re: Qwest, Broadwing Go Gigabit Ethernet I basically agree.
Actually, there's a lot of hype in the ethernet claims. Using Ethernet as an on-ramp (ie, last mile) seems to be an obvious choice. But trying to scale a layer-2 network to metro-sizes is very doubtful (of course, it's do-able in some environments).
But ultimately, that traffic will need to hit a router. If that router is anywhere between the edge-of-the-core and the access edge, Ethernet's advatanges in the Metro area evaporate, and one might as well go Packet-over-SONET.
As it turns out, many service providers claiming to be based on Ethernet only use Ethernet in the last mile (as an access interface) or possibly map ethernet over SONET.
There's a lot of marketing hype in this issue, though I'm sure future posters are going to describe the benefits of ethernet-in-the-metro.

So go ahead...
fiber_r_us 12/4/2012 | 8:27:37 PM
re: Qwest, Broadwing Go Gigabit Ethernet >I'm sure future posters are going to describe
>the benefits of ethernet-in-the-metro.
>So go ahead...

Nah... You have already heard it an know its the right answer. So, no point in covering it again :)
bwuser 12/4/2012 | 8:27:34 PM
re: Qwest, Broadwing Go Gigabit Ethernet Is there any difference between routers and Layer 3 switches? Aren't both devices capable of running routing protocols and routing IP traffic? Cisco seems to have introduced a new router platform (7600) targeted at Metro applications. How is this box different from Layer 3 products from Extreme, Foundry & Riverstone?
Belzebutt 12/4/2012 | 8:27:34 PM
re: Qwest, Broadwing Go Gigabit Ethernet Layer 3 switches are primarily switches with routing functions built on top. They are designed to have a huge fan-out and should be able to handle huge amounts of traffic. However, their routing capabilities aren't usually as good as for routers, they don't necessairly support as many routing protocols and don't support them as well.

For example a Layer 3 switch will not usually support BGP very well (or not at all) and will collapse when you throw too many routes at it. On the other hand, some routers will collapse if you throw too much traffic at line-rate, but they usually won't even support nearly the same fanout as Layer 3 switches.
fiber_r_us 12/4/2012 | 8:27:32 PM
re: Qwest, Broadwing Go Gigabit Ethernet There is no fundamental differnce in terms of basic functionality between the Cisco 7600 (which is really a re-packaged Catalyst 6500) and a Cisco 12000 series. They run the same protocols and the same basic IOS.

Comparing the functionality between Cisco and Extreme, Foundry, Riverstone, Juniper, etc... is a complex task. Indeed, the recent router lab tests so hotly contested on these message boards demonstrates that it is not a simple matter to build a L3-switch/router that could perform well in an Internet environment.

In any event, the provider industry has historically created a very fuzzy dividing line between "L3 switches" and "routers" as they relate to usage as a system in a provider's network:

Interfaces: "Routers" support practically every interface under the sun (10/100/Gig Ethernet, SONET OC-3/12/48/192 with POS, Frame Relay, and ATM encapsualtions, FDDI, Token Ring, VG Anylan, etc). "L3 Switches" have been mostly dominated by Ethernet interfaces with some limited support of other interfaces.

Performance: "Routers" generally attempt to be "line-rate and non-blocking" devices with sufficient interface buffering (100s of MBytes) to make them the most suitable device for "core Internet routing". "L3 switches" tend to be blocking to one extent or another and do not have huge amounts of port-level packet buffers (usually 100s of KBytes).

Protocols: "Routers" support heavy use of feature-rich implementations of BGP, OSPF, IS-IS, MPLS, and other difficult-to-implement protocols. "L3 switches" tend to support some of these protocols, but the implementations tend not to be as scalable or as feature-rich as a true "router". Also, most "routers" have hardware support for forwarding packets no matter how they are encpsulated (MPLS or IP).

Packaging: Many "Routers" are designed to be NEBS compliant and to be placed into carrier facilities (DC power, front-to-back air flow, etc). "L3 Switches" tend to be more designed for Enterprise installations.

Volume: "Routers" are sold by the 10's of thousands. "L3 Switches" are sold by the 100's of thousands. Hence:

Cost: "Router" are expensive. "L3 Switches" are cheaper. This is due mostly to the non-blocking switch fabrics and the amount of buffer memory that exists in routers.

As the industry moves forward, the lines between these two terms will blur even further. Indeed, the recent push for Ethernet everywhere is driven by the fact that bandwidth is becomming cheaper and "L3 Ethernet swtiches" are cheaper than "routers". It follows that big cheap networks can soon be built simply from using Layer 3 switches (with decent protocol software implementations) and sufficient bandwidth such that ports don't have to have huge amounts of packet buffers and the switching fabrics don't have to be line-rate and non-blocking.

The argument basically boils down to:

1) Do you need to build networks out of the absolute peak-performing platforms and squeeze every bit of available bandwidth out of each link, no matter what the cost?


2) Do you build the network out of lots of cheaper platforms that are not built like Formula-1 racing cars and run the links at something less than 90+%, while saving a lot of cash?

If history has anything to say about it, option #2 will win. Lots of cheap, simple, distributed horsepower vs. a few big, expensive, complex, centralized monsters. Seems a lot like the PCs vs Mainframes arguments of the early '80s.

Proponents of option #1 probably still think computing should be done on mainframes. They like the feeling of the "big iron". It a status symbol and ego boost for them ("I have the hotest, biggest, most expensive box in the world").

Proponents of option #2 recognized the power of the PC when it came to computing. They like the feeling of cheap, but adequate, power that belongs to them. This is not an ego trip for them; it is simply a matter of being practical.
gea 12/4/2012 | 8:27:32 PM
re: Qwest, Broadwing Go Gigabit Ethernet Waitaminute...
The biggest difference between a router and a switch is that to pass traffic between LANs, you have to go to Layer 3 and so need a router.

The moment you go through a router, the Layer2 MAC addresses are discarded. Your network is now routed, not switched.

And the moment your network is routed & Layer 3, you ain't running an Ethernet network anymore. You may as well use POS interfaces, because their cost is only marginally more expensive than Ethernet interfaces operated over MAN distances (these are not the same transcievers used inside an enterprise), and because you can now realize the features developed within SONET for runnning optical networks (if you want).

fiber_r_us 12/4/2012 | 8:27:30 PM
re: Qwest, Broadwing Go Gigabit Ethernet The discsussion was about the difference between a Layer 3 switch and a router (which essentially is very little).

Compare the cost of building a network with GSRs and OC-192 interfaces vs building the same network with 7600s (or 6500s) with 10GE interfaces. Which one is cheaper?
skeptic 12/4/2012 | 8:27:29 PM
re: Qwest, Broadwing Go Gigabit Ethernet If history has anything to say about it, option #2 will win. Lots of cheap, simple, distributed horsepower vs. a few big, expensive, complex, centralized monsters. Seems a lot like the PCs vs Mainframes arguments of the early '80s.

What you don't understand is that when your network gets to a certain scale, lots of cheap, "simple", distribute horsepower creates more problems than it solves and in routers, always costs more than a large centralized solution.

PC's and mainframes were both standalone platforms. Routers have interconnects and the more small routers you have, the more money you are going to be wasting on interconnect fiber and line cards among your own equipment. And the bigger the network gets, the bigger the cost in every respect of all those small platforms. The mathmatics of meshes just starts to work against it at a particular point.

The cost of the router isn't the switch or the control plane, its the line cards. Small routers take more line cards than big routers to perform the same task.

And as far as 10GE goes, when I look at the standards as they currently exist, it doesn't look a whole lot like ethernet anymore except at the software level. The components look a whole lot like OC192 and unless the volumes for 10GE are huge, it will not have a significant advantage over sonet in price.

L3 switches and routers each have their place in a network. Neither is going to replace the other in part because the L3 switch manufacturers have to compromise their software and their quality to compete on price.

Routing software on some of these platforms is often an afterthought. The company finds the cheapest software on the open market and throws it on the platform, often without even testing to see if it works.

Router companies are about quality, customers and software (well in theory anyway). L3 switch companies often don't care about software or customers. Often they sit around in meetings with their biggest concern being shaving a few pennies off the cost of a component.
Page 1 / 3   >   >>
Sign In