Ethernet gets active in its pitch against PON

When Verizon Communications, SBC Communications and BellSouth banded together to put out a request for proposal for a massive fiber-to-the-premises build, the debate should have ended. At the core of the voluminous request was an expressly stated desire to deploy a passive optical network, or PON, architecture in which fiber assets are shared across the access portion of the network, with minimal electronics along the way.

Though not in synch with many of the large carriers across the globe, the decision by the three biggest U.S. local operators to use PON should have put to rest the debate over which way most of the country would be receiving next-generation voice, data and video services via fiber. So why are people like Barry Kantner, vice president of marketing for World Wide Packets — a company entirely dependent on selling active Ethernet architectures — so confident these days? Well, without saying the exact words, those in the business of pushing active architectures believe the three RBOCs made a big mistake by selecting PON — and that eventually they will come to realize that to fully utilize the high-bandwidth services that FTTP promises, they must select an active architecture. And while newer versions of PON, like Ethernet PON, allow carriers to offer significantly higher bandwidths than previous generations, the business plan is based on large telcos grabbing large chunks of market share.

“The more subscribers you put on a passive network, the less bandwidth you're delivering to the user,” said Kantner. “This is kind of the dirty little secret of passive optical networks. To be the most cost effective, you have to have the maximum number of users, which is the worst case from a bandwidth perspective. At some point you have so many subscribers, you can't throw any more bandwidth at it.”

Indeed, much of the argument for PON rests largely on economics. In an effort to keep down civil engineering costs (the digging of the trench and actual construction elements), PON shares the most expensive network elements among all users. In areas where telcos want to minimize the amount of fiber deployed, PON architectures call for one fiber to be brought to a neighborhood node or optical line terminal where the signal is split up to 32 ways and sent to each home over another fiber.

In an active Ethernet network, carriers deploy significantly more fiber to neighborhood nodes, run all services over an Ethernet protocol and do an optical-to-electrical-to-optical conversion and include other active elements in the access network. By nature, active Ethernet architectures are point-to-point.

“I think PON came from a need and a desire to minimize the civil engineering costs in fiber to the home,” said Bill Zakowski, vice president of business development for Amedia Networks. “What's changed since then is that the cost of Ethernet has come down. That helped to drive the cost of these Ethernet chipsets to an all-time low. There are benefits of bringing switched Ethernet close to the home, but even moreso when you put it in the aggregation plant. It has [those that selected PON] scratching their heads.”

From a pure capital expense perspective, those on the PON side of the debate claim there is no comparison because point-to-point architectures like active Ethernet require more gear and greater fiber management and real estate costs.

“Fiber cost has come down, but you have to attach transceivers on each side, and for each subscriber you need to have two transceivers,” said Sayeed Rashid, senior marketing manager of access networks division for Alcatel, which is one of the largest FTTP vendors in the world. “In PON, you have N number of transceivers for N subscribers.

“With a point-to-point architecture you have twice the electronics. With passive you don't have any active electronics in the outside plant. When ILECs look at it, the business case looks considerably better because you have a passive plant. With active elements you also have to house the electronics somewhere, which means big cabinets — some of which have to be air conditioned because not all of the elements are hardened.”

Perhaps even more important to PON advocates are the operational cost savings that come with removing active elements from the access network. One of PON's biggest marketing points is the ability to put fiber in the ground and leave it alone for years.

“In deploying PON, carriers can reduce their field service head count,” said Dave Lee, vice president of marketing and services for Telco Systems.

For incumbent carriers, ATM-based PON, or APON, also allows them to take advantage of their existing ATM switching infrastructure.

“They're looking at the PON as not only a mature solution, but also a solution that fits their architecture and provides them with carrier-grade functionality,” said Aviv Ronai, associate vice president of ECI's Broadband Access Division, which has been marketing PON mostly in Europe and Asia. “Outside North America, we thought we'd see this change of technologies from ATM to Ethernet. Now we don't see it, though. We see many other operators in Europe discussing PON and seeing PON positioning itself as a telco technology, while Ethernet is being positioned as a CLEC technology.”

However, active Ethernet vendors say comparing costs between the architectures is difficult at best because of the multiplicity of options. Depending on the type of services carriers plan on deploying, the cost of customer premises equipment could radically change. Additionally, there are some vagaries closer to the core that not all PON vendors factor into the overall cost.

“In the PON solution, there's this cloud in the central office. They say ‘Drop in your core switch here,’ but they don't include that in the cost,” said Zakowski. “The electronics for the active solution simply come out cheaper.”

Ethernet's development as a standard in the enterprise over the last 20 years has allowed carriers and municipalities to take advantage of the natural price decline that comes with volume deployments. Utopia, which deployed a nearly state-wide fiber network in Utah, opted to deploy active Ethernet because it found virtually no cost difference, said Sathya Narayanaswamy, director of product marketing for Riverstone, which is providing the project with routing technology.

“When you factor everything into a real-life network, you find that active Ethernet is the same price as PON. There's also a cost to be locked into a single vendor, which can happen on the PON side because it's not really standards-based.”

Added Telco Systems' Lee, “Even though APON is a standard, much of the vendors' equipment has a lot of difference. With Ethernet, pretty much everybody supports the same major tool set.”

Regardless of cost, what ultimately matters to users — and perhaps to some forward-thinking carriers — is the type of services and amount of bandwidth that either architecture can deliver. In Verizon's initial FTTP deployment in Keller, Texas, the company is delivering up to 30 Mb/s to each user, which is lower than the existing equipment can provide, according to several vendors. A handful of independent telcos that have deployed PON architectures are pushing out up to 30 Mb/s to the user, but splitting it up among various services, including bandwidth-hogging video. In Japan, carriers are proving that both PON and active Ethernet can deliver massive amounts of raw bandwidth.

“If you look at the bandwidth that is being delivered via PON systems, NTT is using PON to deliver 100 Mb/s to the customer,” Alcatel's Rashid said. “At the same time NTT East is using point-to-point and they're doing 100 Mb/s. Bandwidth is not just a function of the technology. It's also how you architect the network.”

Using gigabit PON allows carriers to get up to the highest end of Ethernet speeds, though Rashid believes the gating factor to wide-scale deployments beyond 100 Mb/s is the cost of CPE. “I don't think PON by its nature is constrained in bandwidth. We're looking at PON where bandwidth is divided in the optical domain. We're talking about two gigs of bandwidth per subscriber sustained.”

While vendors will continue to quibble over whose bandwidth is bigger, ultimately carriers must decide on the type of services they want to deploy. For FTTP or even FTTN, there is really only one service driving the need: video. The questions now are what kind of video and how much bandwidth it will require.

In the initial implementation of all three RBOCs' video strategies, the joint RFP requested the ability to provide RF video, which is the same format being used by the cable industry. Though many in the IP video world believe it's a step backward technologically, there are several economic reasons for the decision.

In a competitive market where FTTP providers likely will be facing some form of facilities-based competition, there must be some differentiation in services.

“Smart services are not just selling 100 Mb/s,” said Narayanaswamy. “They're saying ‘We'll give you two [high-definition] channels and six VoIP lines.’ That's the way to sell smart services.”

Those on the active Ethernet side tell a different story, one based on massive bandwidth combined with intelligent transport. “Even with G-PON, you still only have about 60 Mb/s to play with,” Kantner said. “A couple of HDTV programs at 19 Mb/s and you're almost out of bandwidth.”

Whether that process will get the largest carriers to dump PON in favor of active Ethernet is still debatable. AFC Communications, which is the primary supplier to Verizon's early FTTP rollouts, isn't talking specifics on any current or future contracts, but according to at least one source within the company, BellSouth and Verizon are exploring all architectures.

Telco Systems president Larry Asten also is among a growing group of industry executives who believe PON shortcomings will eventually force incumbent carriers to reconsider.

“You've got a lot of the incumbents looking at this saying ‘Well, should we put out another RFP,’” Asten said.

Conversely, those on the PON side of the debate claim they have a number of customers that have flip-flopped on the issue back to PON.

“We've been in discussion with multiple operators who have been on the point-to-point page and are moving back,” said ECI's Ronai.

However, at least one large telco is seriously considering the move from PON to active Ethernet. SureWest Communications, which acquired a PON-based FTTH network around Sacramento when it bought WinFirst out of bankruptcy, wants to begin deploying active Ethernet soon, according to Bill DeMuth, chief technology officer for SureWest.

“Some of the active stuff is where we want to get,” he said. “From everything we've seen we can do a lot more with that technology.”

access technology

Visit Telephony's Web site for a story on the ongoing service provider debate over deploying IP video or using an RF overlay to offer video services.
WWW.TELEPHONYONLINE.COM