Back to the Future

PDH Meets Carrier Ethernet

They say you can’t teach an old dog new tricks, but telecom engineers are doing just that by turning their old PDH networks into carriers for carrier Ethernet services. The idea of using T1/T3 circuits to carry Ethernet in the WAN is not entirely new; its use started in earnest about 18 months ago with the introduction of some enabling gear. And, since it first hit the scene, Ethernet over PDH (EoPDH) has found a ready market in service providers trying to keep up with the growing demand for Ethernet services by businesses, for mobile backhaul and access aggregation. In that short time, EoPDH also has evolved to a next generation that uses protocols developed to deliver Ethernet-over-SONET/SDH networks.

“What we are looking at is the last gasp,” says analyst James Heath, research director for broadband at Dittberner Associates Inc. “We have PDH networks in place, and people are trying to roll out new broadband services and new wireless services very quickly. They don’t have the time or money or personnel in a lot of instances to do everything. What they are trying to do is get new services deployed on an old infrastructure.”

PDH connections remain plentiful. For instance, more than 70 percent of all mobile backhaul connections (copper and radio) worldwide are PDH-based, according to Infonetics Research. And, according to research from Vertical Systems Group Inc., only 13.4 percent of buildings are connected with fiber, leaving a majority with copper-based T1/T3 connections.

“There is going to be a lot of legacy around for quite a while,” notes Daniel Joseph Barry, director of marketing for Tpack. “If you have a PDH connection as well, just reuse it. There is no reason that you can’t reuse it until the bandwidth grows to a point that you have to get rid of it.”

Extending Ethernet from LAN to WAN may be a driver for EoPDH, but there already were ways to move Ethernet from one location to another using routers and Layer 3 transmission. “The bottom line is that it’s much easier and much less expensive to operate at Layer 2 if you can get away with it,” says Larry Jacobs, vice president of marketing, RAD Data Communications Ltd. “That is what has been driving all this buzz you are hearing about.”

There was a growing buzz over EoPDH at this year’s biggest industry event, NXTcomm, which was held in late June in Chicago. At the show, there were numerous product announcements from vendors, such as Alcatel-Lucent, Nortel Networks Ltd., RAD and OEMs like Telrad Networks Ltd. and Tpack.

RAD got creative with one offer, the MiRICi-T1/T3, which puts an EoPDH bridge into a small form-factor pluggable package normally used for compact optical transceivers. This enables EoPDH to be added into installed routers and switches as needed.

RAD also introduced a more traditional EoPDH box, the RICi-16, which is a combination Ethernet/ DS1/DS3 multiplexer that utilizes circuit bonding to create a virtual large pipe composed of four, eight or 16 T1 lines to deliver high-speed Ethernet.

“The rule of thumb in the industry is that it is not economical to buy more than five or six T1s, that once you do that, it’s cheaper to buy a T3,” says Jacobs. “The trouble is that many [cell] towers and other locations don’t have coax running to them that you need to do T3 over, but they do have copper. So, even though the economics may be more favorable to T3s if you had the choice, in many situations you don’t.”

Where there are T3s, such as in a cellular hub location, the RICi-16 can be divided and allocated to deliver Ethernet or voice over T1s. “They could take a DS3, which is 28 T1s, and assign 10 T1s to backhaul and the remaining 18 for Ethernet. Now, you might have a pipe of 30MB of Ethernet and 10 T1s,” says Eitan Schwartz, vice president of pseudowire and Ethernet access for RAD. “Over time, they will require less and less T1s. They can just reallocate bandwidth on an as-needed basis from TDM to Ethernet.”

Both RAD products support generic framing procedure (GFP), which has been used to map client signals like Ethernet-over-SONET/SDH networks.

Previously, vendors used other protocols, such as inverse multiplexing for ATM (IMA) and multilink point-to-point protocol (ML-PPP) for link aggregation as well as encapsulation techniques like High-Level Data Link Control (HDLC) to transport packet data over PDH connections.

According to a white paper published in May by Tpack and TranSwitch Corp., the newer approach uses GFP to encapsulate the packet data, VCAT to concatenate the PDH connections for transmission of packet data, and LCAS to dynamically add or take away PDH capacity. The main advantage of this approach is “fewer dropped packets, predictable low overhead and up to 20 percent greater throughput,” the report claims. GFP, the paper explains, has mechanisms for reducing the number of packet retransmissions due to errors, and it also has a fixed frame overhead for predictable bandwidth. In the event of failures, VCAT/LCAS provide for hitless restoration of traffic.

“What LCAS gives you if you have [several T1s] bonded together, if one goes down, the whole traffic doesn’t go down,” says Vijay Malik, vice president of product architecture for RAD. “It basically allows you to expand and contract the bandwidth that is in your transport network. This bandwidth can be added or deleted based on customer requirements or failure in the network.”

When compared with IMA and ML-PPP, GFP/VCAT/LCAS offers “superior latency characteristics, which can be an advantage for time-sensitive services, such as real-time video,” according to the Tpack/TranSwitch white paper. (See table below comparing GFP/VCAT/LCAS with PPP/HDLC/ML-PPP).

“You have technical solutions [for EoPDH] — IMA or ML-PPP. The beauty is in Ethernet,” says Tpack’s Barry. “Once you introduce that, at anytime you want to change it to fiber or a broader radio, the Ethernet interface stays the same. It’s building for the future as well.”

In addition, using Ethernet-based protocols instead of the ATM-based IMA or IP-based PPP, carriers can use the User Network Interface (UNI) defined by the Metro Ethernet Forum (MEF) so that compliance with performance parameters, such as jitter and packet loss, can be tested. Adopting the MEF UNI also allows a planned transition to a packet-based infrastructure, according to the Tpack/TranSwitch white paper.

TranSwitch and Tpack have introduced an end-to-end EoPDH solution for OEMs that combines the TranSwitch EtherMap-PDH device for mapping Ethernet date to PDH and Tpack’s Ethernet-over-PDH/SONET/SDH packet mapper for aggregation and processing of multiple Ethernet-over-PDH connections. No OEM deals have been announced yet.

TranSwitch also works with Telrad on its TAG-10 solution, which was announced for sale to OEMs at NXTcomm. The offer includes two pieces: the TAG-10a standards-compliant AMC card for use at the aggregation end and the TAG-10c CPE for use at the edge. The TAG-10 solution enables Layer 1 mapping of Ethernet frames using standard GFP/HDLC/LAPS framing into bonded DS1/E1/J1 connections.

Rebecca Rachmany, Telrad’s director of marketing, says OEMs are turning to companies like Telrad for niche applications, such as TAG-10, which is designed to support hybrid networks. “The R&D is quite large compared to the [addressable] market,” she says. “For us, since it’s something we specialize in, we can reduce the costs and time-to-market.”

At NXTcomm, OEM Nortel Networks announced its Optical Multiservice Edge (OME) 6500 PDH Gateway interface, which is a card for the OME 6500 that can subtend CPE to offer Ethernet over bonded T1s or DS3s. Nortel also announced an agreement with ANDA Networks Inc. to bundle and resell ANDA’s GFP-based EtherReach 2000 EoTDM CPE devices along with the OME 6500 with the PDH Gateway. “[The package] cost-effectively allows [service providers] to offer that Ethernet service and flexible bandwidth rates and consistent SLAs with their other Ethernet services over their existing copper-based network infrastructure,” says Scott McFeely, vice president for Metro Ethernet Networks product line management, Nortel. “In other words, it allows them to extend their Ethernet service offering to places where they may not necessarily have the economics or the time to get to a fiber upgrade.”

Alcatel-Lucent’s offer in the space is the TSS 1850-3, a version of its 1850 Transport Service Switch (TSS) that switches up to 3gbps. Announced at NXTcomm, this customer-located equipment chassis fits into hybrid Ethernet and TDM access networks providing a cost-effective solution for Ethernet in the first mile service demarcation between a carrier network and the end-user and Ethernet aggregation. It comes in multiple configurations, including one with PDH uplinks.

RAD Packs EoPDH System Into SFP

RAD Data Communications Ltd. has put an entire Ethernet-over-PDH solution on a small form pluggable (SFP), eliminating the need for an external device, reducing costs and increasing network flexibility. The finger-sized MiRICi-T1/T3 is housed in a standard SFP package, so that it can be plugged into any Ethernet switch with compatible sockets to provide remote bridging over T1 and T3 lines.

“What RAD did is we took the [SFP] concept and rather than a fiber interface, we built in a bridge to allow you to run Ethernet over T1 or T3 trunks,” says Eitan Schwartz, vice president of pseudowire and Ethernet access for RAD. “You can just plug it into a switch or router or other Ethernet-based device.”

The MiRICi-T1/T3 works with any vendor’s gear that has MSA-compliant sockets. RAD has a few options of its own, including its ETX Ethernet demarcation and aggregation family, its IPMUX pseudowire family and its ACE ATM family of products. The MiRICi-T1/T3 draws power from the host device, but does not require special software since its onboard ASIC permits it to function autonomously. It includes full OAM capabilities. The MiRICi-T1/T3 supports GFP encapsulation for interoperability with third-party GFP-compliant devices as well as bookend connectivity.

The SFP costs around $500 depending on the interface, Schwartz says. “It’s much less expensive than a standalone box,” he says, noting that adding a T3 interface on an existing router or switch can be expensive and there may not be an option to buy the number you need, forcing you to purchase additional gear. “Buying a switch with all the right physical interfaces is going to be incredibly expensive. Buying a switch that just has SFP sockets and then just plugging in whatever makes sense for a specific customer gives them a lot of flexibility.”

Additionally, the SFP approach enables service providers to change out the interface as needed. “Maybe six months down the road, the customer needs have grown and they can warrant delivery of fiber or maybe they were in the middle of construction and it’s completed and they can deliver fiber. They can unplug the [MiRICi-T1/T3] SFP and plug in a fiber SFP,” Schwartz explains. “It gives them that future-proof capability.”

Share this article: Email, Slashdot, Digg, Del.icio.us, Yahoo!MyWeb, Windows Live Favorites, Furl
Add this article feed to: RSS, My Yahoo, Newsgator, Bloglines