June 1, 2011
The New Straight Talk On DPoE
By George Lawton
Today, MSOs use two distinct management systems for their traditional hybrid fiber coax (HFC) networks and for their new fiber-optic EPON networks. The new CableLabs spec promises to bring these two worlds together.
The DOCSIS Provisioning of EPON (DPoE) specification took a major leap forward with the release of DPoE v1.0 in March. At its heart, the standard promises to make it easier to bill, manage and invoice new fiber-optic based services using DOCSIS tools.
MSOs traditionally have deployed two distinct management systems for their traditional hybrid fiber coax (HFC) networks and for their new fiber-optic EPON networks. CableLabs’ new spec promises to bring these two worlds together.
“EPON and DPoE don't really come into play until cable operators start asking about what comes after DOCSIS 3.0.” — Jeffrey Heynen, Infonetics Research
“The DPoE specifications will allow cable operators to deploy EPON equipment for business services and use the same back office servers used by DOCSIS networks to provision the equipment,” says Robert Harris, vice president/Network Planning and Architecture at Time Warner Cable.
Version 1.0 of the DPoE specifications defines the requirements and functionality to support Metro Ethernet Forum (MEF) Ethernet Private Line (EPL) service for business customers. Thirteen vendors have participated in the interoperability demos held at CableLabs, including Adva, Arris, Atheros, Broadcom, Ciena, CommScope, CTDI, Finisar, Huawei, MRV, Sumitomo, Telco Systems and ZTE.
DPoE addresses one of the biggest challenges MSOs have in scaling up fiber-optic-based commercial services. Shane Eleniak, vice president/Advanced Broadband Solutions at CommScope, explains, “It does not matter whether you are SBC or Comcast; if you bring something new into your network, it can get expensive if you have to make changes to the billing and management system. With DPoE, you will get to use EPON the same way you deliver DOCSIS services today.”
Jeffrey Heynen, senior analyst with Infonetics Research, says MSOs now are deploying some 4,000 new EPON services per year. The scale of cutover is small enough and the use case different enough that cablecos can use existing SNMP management tools rather than DOCSIS. Over time, cable operators gradually will be able to transition the DOCSIS subscribers to fiber. “Ultimately, it is a cap-and-grow architecture rather than rip-and-replace,” he explains.
Where It Fits
Today, most MSO services are provisioned using DOCSIS, but EPON equipment has been built on top of the SNMP management protocol. Consequently, a distinct new set of cable-technician skill sets is required for EPON customers.
“There is some thinking by executives that they would rather do everything with DOCSIS, and that it would be better to have one management system than two,” notes Mannix O'Connor, former director of marketing at Hitachi. “They are thinking about a few years out, when they have more fiber services.”
MSOs and telcos have developed different management-system architectures. MSOs have focused on DOCSIS management systems, while telcos have focused on TL1, which is better suited for their circuit-switched architecture.
Most telcos that have deployed PONs have chosen GPON because of its better compatibility with circuit-switched network architecture, driven by Verizon, SBC and British Telecom. The traffic container (Tcont) in GPON provides a virtual circuit container for all of the traffic. This mimics the virtual circuit system commonly used by telco management systems.
Telco and wireless applications also have had a strong need for process-time-synchronization data, which traditionally has been available over telephone circuits, but not over Ethernet. However, the recent IEEE 1588 v2 precision time protocol will extend this capability to Ethernet.
Cable providers have tended to adopt EPON because of their greater familiarity with Ethernet. EPON uses SNMP to do the management, which has not been compatible with DOCSIS. According to Infonetics’ Heynen, there is no inherent advantage to either GPON or EPON at the moment aside from architectural compatibility.
SNMP and DOCSIS are relatively straightforward to integrate because they use a similar point to multi-point architecture. CommScope’s Eleniak says, “MSOs are used to seeing IP-based traffic, and then they use DOCSIS as their native protocol.”
EPON also looks similar to DOCSIS in terms of traffic classes — the way it is carried and the type of queues. For example, DOCSIS cable-modem ports are called SSIDs (Service Set) in contrast with LLIDs (Link Level) used on EPON.
The main difference between DOCSIS and SNMP is the management model, Eleniak adds. With DOCSIS, a cable modem registers with DHCP, receives an IP address and uses Trivial File Transfer Protocol (TFTP) to pull down a configuration file. With SNMP, the customer equipment retrieves several Management Information Base (MIB) commands for setting the correct parameters. DPoE is designed to mimic the configurations of CMTS and cable modems, and to act as a proxy between the DOCSIS management world and SNMP.
“In DOCSIS, there are a whole bunch of DOCSIS MIBs with type values. With EPON, we wanted a standard set of SNMP MIBs so that every vendor's DPoE would support the same set of MIBs with a consistent set of message between then,” Eleniak explains. “This allows the MSO to leverage all of its fault-management and configuration-management-system tools to offer IP over DOCSIS, but only now invoice and bill a customer using the same tool. DPoE basically lets MSOs leverage their existing back offices.”
“DPoE 1.0 focused on single-port devices, single-port services and Ethernet virtual-private-line services. Version 2.0 could include multi-port services and devices.”
DPoE 1.0 focused on single-port devices, single-port services and Ethernet virtual-private-line services. Version 2.0, still in development, could include multi-port services and devices, VoIP and point-to-multipoint. Plans call for testing against these capabilities, starting in August.
One of the benefits of DPoE is that it allows interoperability of equipment from different vendors. Traditionally, MSOs have had to buy matched sets of Optical Network Terminals (ONTs) and Optical Line Terminals (OLTs). DPoE will allow MSOs to buy the best ONT, OLT and management pieces, and to create the best services.
In parallel with these efforts, IEEE P1904.1 is doing related work on Service Interoperability over EPON (SIEPON). The DPoE and SIEPON groups are exchanging ideas. CableLabs has hosted the SIEPON working group, which is focused on adding broader management features for all kinds of service providers, rather than just focusing on the needs of MSOs.
The main pushback to the wider adoption of EPON has been the cost of fiber, Heynen says, adding telcos have been spending money to upgrade to DOCSIS 3.0, which provides sufficient bandwidth for the short run.
He continues, “EPON and DPoE don't really come into play until cable operators start asking about what comes after DOCSIS 3.0.”
There are other challenges relating to educating installers. O'Connor notes, “Some of the people in the field that install services don't want to use DOCSIS. They see it as a limiting form factor and less capable than the software they use today. But that is because they haven't seen the new DPoE specs out of CableLabs.”
For example, a cable modem only can be used by one customer; as such, a cableco would have to install 5 cable modems in a building to reach five customers because DOCSIS does not have a way to put multiple customers on one modem. With Ethernet, the provider can install one box in the building and then add new customers via software. O'Connor adds, “They think of DOCSIS-based devices as having the same limitations as cable modems, but this does not have to be the case.”
Hitachi has been a key player in DPoE efforts through its Salira subsidiary. However, the company recently left the U.S. market entirely due to slow growth and to the recent earthquake in the Pacific Rim. “DPoE is viable if you are a company selling into cable with a big installed base rather than a niche product,” he explains. “A pure DPoE chassis does not have a good market. While none of the vendors have DPoE, some will integrate it into their CMTS chassis in the future.”
The Road Ahead
CommScope’s Eleniak sees three main markets for DPoE: serving large companies; wireless backhaul; and for greenfield residential services in Europe, South America and Canada. However, he does not expect strong residential penetration in the United States. In the past, he says, most fiber optics only were used in active systems for the largest customers. However, the rise of PONs promises to reduce the amount of fiber required to deploy a network, which could help drive DPoE into business applications first. He expects to see the first RFPs for DPoE later this year.
Many MSOs continue to wait. Eleniak noted, “A lot of the larger MSO don't see a need to go to something like this. DOCSIS, 3.0, switched digital video and analog reclamation will allow them to deliver all the bandwidth they need for the foreseeable future.”
Analyst Heynen does not expect to see DPoE take off in U.S. residential markets for another five years, owing to the costs of transitioning set-top boxes.
“No one is talking about IP-based set-top boxes but, ultimately, cable operators want to move to IP-based video,” he says. “The key is moving from MPEG-2 transport to IP transport.”
O'Connor expects to see DPoE deployed in packets across the United States, but he thinks spending for fiber might compete with other new technologies like ad insertion, and the margins might not be as high.
–Technical writer George Lawton is a frequent contributor to Communications Technology. Contact him at firstname.lastname@example.org