To create an IP/HFC network "on-button," cable operators will have to focus on some core network elements.
Imagine this scenario. A cable operator is able to aggregate all of its Internet protocol (IP)-based voice, video and data traffic from its residential and business subscribers into a single IP aggregation switch at the local distribution hub or headend. The traffic from this and other IP aggregation switches all over the metro network are consolidated into redundant regional edge routers, which communicate with a national backbone controlled by the cable operator.
Content sources, such as Internet uplinks, voice gateways and video-on-demand (VOD) servers, are all accessible through this backbone. The network devices and transport layer that make up the metro network are able to route and switch next-generation services like voice-over-IP (VoIP) and virtual private networks (VPNs) with carrier-class performance and five nines availability. It is the MSO’s dream. The operator carries all of its own traffic, never having to hand it off at any point to incumbent local exchange carriers (ILECs) or other service providers.
Laying the groundwork
The dream isn’t that far-fetched. Today’s cable operators can deliver just about any next-generation service they want, including voice. Cable operators’ existing hybrid fiber/coax (HFC) networks, already upgraded to deploy data services by cable modem, can support a wider berth of IP traffic than the telcos’ copper plant.
Data Over Cable Service Interface Specification (DOCSIS) has played a key role in enabling that capability, first by affording the HFC network the ability to support two-way (primarily data) traffic for residential cable modems, and then by addressing quality of service (QoS) issues such as latency control and guaranteed service levels. The newest version, DOCSIS 2.0, will increase the bandwidth and throughput of upstream transmissions, giving an even greater boost to performance-reliant services such as voice and videoconferencing. Another major benefit is that the current crop of cable modems sitting today in subscribers’ homes and businesses are perfectly capable of handling next-generation IP services without need for further modification.
For the previously mentioned dream scenario to work, however, all of the traffic riding over operators’ networks must be IP-based, and except for the vanilla high-speed access available through today’s cable modems, next-gen services currently are not carried as packetized, IP traffic. Voice, for instance, is in most cases carried over separate, stand-alone HFC telephony equipment. That’s because, despite the evolution of DOCSIS and some key network technologies, several more elements must be in place to make the all-IP dream a reality.
Beyond performance, operators still face the challenge of availability and redundancy. Their transport technology must be able to support the huge volume of IP traffic generated by next-gen services. And their equipment must take on privacy and security functions that much of today’s technology lacks. (Operational support and business support systems are crucial, as well, but this article will focus strictly on the network.)
Offering viable residential and commercial IP services means achieving carrier-class "five nines" availability and performance. The answer to attaining them lies in having the right technologies in place, and solving key network configuration issues. This article considers some of these elements that are most vital for cable operators to master.
Edge devices: The CMTS
The all-IP network begins at the edge. While every segment of network topology is important, none performs more functions in an all-IP environment than the cable modem termination system (CMTS)—the headend/hub device closest to the subscriber.
The good news is that, in contrast to the first generation of CMTSs, today’s devices have a much higher degree of IP functionality. First-generation CMTSs merely supported high-speed Internet access, which meant either low-performance routing or bridging. Now the CMTS acts as a true edge router—the clearinghouse for all packets going to and from the subscribers’ modems or home routers.
As such, the CMTS integrates the functions of many devices, including wire-speed IP routing, service provisioning, RF upconverters, network diagnostics and various interface modules. The integrated device makes for more robust connectivity, and cable operators manage one device versus many. Even with so many subscriber-side elements in play, operators can query just one device, the CMTS, to make maintenance decisions.
What’s more, the configuration of the CMTS addresses IP-related availability issues. Where first-generation CMTSs lacked redundant management, DOCSIS and route processing modules that are vital to uninterrupted packet transfer, today’s products feature fully redundant architectures. (See Figure 1)
The CMTS plays a major role in ensuring end-to-end quality of service (QoS). Because of the great disparity in QoS functionality in the different metro and backbone Layer 2 technologies such as asynchronous transfer mode (ATM), packet-over-SONET (POS) and gigabit Ethernet (GigE), the only QoS features that are available end-to-end in these networks are those provided at the IP layer by protocols such as type of service (TOS), DiffServ and multiprotocol label switching-traffic engineering (MPLS-TE). Because the CMTS resides at the junction point of DOCSIS and the traditional IP network, it successfully must marry DOCSIS and IP QoS capabilities.
In addition to supporting a full suite of IP routing features, next-generation CMTSs also must integrate the security features required in a public network edge router. These features, such as proxy-address resolution protocol (ARP), dynamic host configuration protocol (DHCP) lease query, broadcast suppression, DHCP authority and others are essential to guarantee IP connectivity to normal subscribers who do not attempt to bypass the DHCP infrastructure. These features prevent rogue subscribers from spoofing other subscribers’ IP and MAC addresses, two of the most common methods used to launch denial of service and other forms of attacks.
Up to now, cable operators have placed an edge router with their CMTS as the interface to the metro network. With next-generation CMTSs and their routing functionality, this is no longer necessary. More operators are replacing edge routers with cheaper and faster Layer 2 aggregation switches, and are using low-cost GigE interfaces to aggregate all of the traffic from their CMTSs into this device.
Having a Layer 2 switch has two benefits. The first is performance. It is simpler and much faster to forward MAC information than IP packets. The second is cost. Layer 2 devices are simple to configure and much cheaper to build.
Edge routers still play a role at the regional distribution hub or headend, collecting traffic from all the aggregation switches on the metro network and offloading it to the cable operator’s backbone network. Even at the headend, however, aggregation switches would be useful for supporting a variety of servers used for DHCP control, web caching, email, news and gaming. (See Figure 2)
At the distribution hub, the CMTS and the local aggregation switch provide nearly everything a cable operator needs to enable IP services. The next step is making sure that the transport layer supports quick and effective transmission of the gigabit traffic created by these services.
Traditionally, operators have relied on ATM and POS. Both are proven and widely in use, but they are showing their age. ATM’s main drawbacks are its configuration, management complexity and high cost. In addition, ATM’s cell-based transmission was designed to handle voice, video and data traffic together, in a single stream. ATM’s performance, though very good in a permanent circuit, diminishes in a "many-to-many" IP environment. It is at its best when handling traffic on its own—without the IP layer.
Performance is not so much the issue with packet-over-SONET as it is a clean, simple network topology. The major drawback with POS is that it requires point-to-point links between distribution hubs. This means that operators must connect each and every one of their hubs in a one-to-one enmeshed configuration of pipes, with a dedicated interface in each local aggregation device, or they must connect their buildings in a ring that forces the Layer 2 switch to actively forward every packet transmitted over the ring. While operators would have the benefit of self-healing fiber rings, they would nevertheless face greater network inefficiency and cost.
Finding lower cost solutions
Operators that rely on ATM or POS have two choices: keep extending the increasingly complex legacy plant, or cap investments made in these technologies and invest in faster, lower-cost technologies. Operators can continue to choose the first option, but eventually, they will have to enact the second.
While POS can work well in a ring configuration with Layer 2 devices as "hops," a more effective method is to create a ring using gigabit Ethernet over fiber. (See Figure 3)
GigE, as it is called, is the best option available, because it provides a greater amount of long-haul bandwidth, which can be vastly expanded with dense wavelength division multiplexing (DWDM). Though it lacks SONET’s quick self-healing characteristics (the open shortest path first (OSPF) protocol provides slower-moving convergence), GigE is much cheaper and faster, and requires no special gear.
The coming of RPR
For today’s cable operators, GigE is the most popular answer, but an interface now in the process of being standardized may well lead the way to the all-IP network envisioned at the beginning of this article.
Resilient packet ring (RPR) will enable operators to connect metro networks through cheap, resilient and easy-to-manage fiber rings. The technology serves as an excellent interface because it directly interconnects all the aggregation switches over a common shared-bandwidth ring, allowing for spatial reuse, ring protection and fast restoration. It is in this configuration that all of the aggregation switches in each distribution hub aggregate to a single edge router. IP traffic flows across the ring, uninhibited, from a Layer 2 device to the edge router, without being delayed by Layer 3 devices. The result is a faster, more efficient and streamlined methodology for transporting packetized traffic.
Putting it all together
Once a cable operator has upgraded its distribution hubs and streamlined its metro networks, it can replicate much of its work on a national level, using the same techniques to ring all of its properties and creating one or a few regional network rings out of many local ones. In the case of a national backbone, the transport technology of choice may still be POS, because the jury is still out on whether other technologies can match POS for distance and reliability.
In summary, operators must start at the network edge to enable full IP routing capabilities, with a full array of protocols to ensure end-to-end QoS. Although edge routers were previously the most popular approach to aggregate CMTS traffic into the metro network, many operators worldwide are turning to faster and cheaper Layer 2 aggregation switches for this function. This is possible because of the expanded IP feature set found in next-generation CMTSs. In addition, cable operators must account for security and redundancy in the right places, primarily at the subscriber edge. New transport technologies such as GigE/DWDM and RPR help further reduce the cost, while increasing the available bandwidth of regional networks. With these technologies in place, cable operators can realize the dream of the true IP/HFC network. How quickly it happens is up to them.
Benoit Legault is the director of ADC’s IP Cable Technology Consulting Group in Westborough, Mass. Email him at: [email protected]
IP Over HFC: Maximize Your Plant
Cable operators can leverage their existing HFC plant—already upgraded for cable modem services—to support a wider berth of IP traffic than the telcos’ copper plant. They must start at the network edge to enable full IP routing capabilities, with a full array of protocols to ensure end-to-end QoS.
Although edge routers were previously the most popular approach to aggregate CMTS traffic into the metro network, many operators worldwide are turning to faster and cheaper Layer 2 aggregation switches for this function. This is possible because of the expanded IP feature set found in next- generation CMTSs.
In addition, cable operators must account for security and redundancy in the right places, primarily at the subscriber edge. New transport technologies such as GigE/DWDM and RPR help further reduce the cost, while increasing the available bandwidth of regional networks. With these technologies in place, cable operators can realize the dream of the true IP/HFC network.