Eye on the Plant

Peer to peer (P2P) file-sharing applications have challenged cable’s data networks ever since their round-the-clock appetite for bandwidth disrupted business models that had assumed different patterns of subscriber behavior. The applications themselves have been fast-moving, evolutionary targets, with the "super-node" model surviving as the fittest. That hierarchical approach to P2P is most often associated with Kazaa. With 375 million reported downloads, Kazaa’s proprietary software has nonetheless declined in popularity in the face of competition from open-standard rivals. Kazaa co-founder and original CEO Niklas Zennstrom himself adopted the super-node approach for telephony through Skype, the service he launched in August 2003. Opportunistic app Since then, Skype has been downloaded more than 25 million times and enabled 1.8 billion minutes of voice service. In October, it was adding 70,000 registered users a day, up from 30,000 in May, and 15,000 in January, Zennstrom says. Speaking via videoconference to the Internet Telephony Conference and Expo in Los Angeles in October, Zennstrom announced Skype’s latest opportunistic play: a business market initiative planned for next year. The initiative itself is demand driven, arising from survey data indicating that close to half of all Skype users had already used their Skype application for business communications. This raises a poignant contrast. Any effort on cable’s part to delivery voice over Internet protocol (VoIP) to businesses is stymied—at least for the time being—by the "fax problem." (See related story, page 12.) Yesterday’s technology poses no obstacle to Skype. An end-to-end PacketCable solution, of course, differs from the individual and workgroup applications that Zennstrom has in mind. But for Skype to have already established a beachhead in the "mission critical" business arena is significant. After all, it’s quality of service (QoS) that causes many of cable’s data engineers to look askance at the P2P voice phenomenon. Define quality Zennstrom believes the supernode P2P model yields a superior voice product. "We’ve been able to set up the communications stream directly between the end users, using the routing on the Internet, rather than routing the calls through some centralized softswitch or server or gateway," he says. Cable’s efforts to meet exacting standards are largely superfluous, he contends. "They are controlling this line, so that if they want to, they can, of course, put QoS mechanisms or protocols over the IP, over the access line," he says. "My experience has been that as long as you have a good broadband connection, there’s no need for QoS mechanisms because you get QoS with capacity." Does mean opinion scoring or other such tools define Skype’s quality? "It’s more like we have 11.9 million users," Zennstrom says. "I think that’s kind of our metrics." The company already has begun adding functionality, such as Skype Out, which enables calls to the public switched telephone network (PSTN). Skype’s original service was free, but for intra-Skype use only. Skype Out users, who number around 200,000, can call within and to 22 different countries for about 2 cents per minute. A quick look at Skype-related Blogs, however, indicates that the quality perceived by this smaller population is at times problematic. QOS-aware network Linking service with physical infrastructure, cable operators have taken pains to optimize their data networks. Cox Communications, for instance, uses modeling tools from OpNet to shift traffic away from transit links and toward settlement-free peering links and intrusion detection and mitigation technology from Arbor Networks. Applications by themselves are vulnerable. "If you don’t have a QoS-aware network, you’re exposed to degradation, particuarly in failure scenarios," Cox Communications VP of Telephone and Data Engineering Jay Rolls says. Congestion, after all, happens. And under bandwidth constraints, quality will suffer, says Benoit Legault, vice president of marketing at Ellacoya Networks. Why? "Because the network will not be able to necessarily prioritize Skype traffic over other types of traffic," he says. In other words, best-effort is just that. Zennstrom says he welcomes the opportunity to partner with broadband companies and cable operators, but any such deals are likely to turn on the question of just how good is best. Jonathan Tombes Systematic, Proactive and Targeted Maintenance As cable continues to expand its services portfolio, preventive maintenance stands as a kind of proxy on underlying network quality. Consider the industry’s past. "The historical practice of cable operators has been to allow sometimes major and often minimal interruptions of service at any time it was convenient for the technician to do it," says Keith Hayes, VP of network operations at Adelphia. The industry’s ongoing shift toward maintenance windows that minimize signal-interrupting repairs on the network has made life less convenient for technicians. Operationally speaking, it has meant sending out teams between 2 and 5 AM, at least to those nodes on the plant not over-populated by enclaves of Web-surfing night owls. Along with cultivating techs with night-owl habits of their own, operators simultaneously have had to acquire eagle-eye vision. "The key to being able to divert most of your maintenance into those windows is to have visibility into the performance of your network, so that you can schedule repairs before they get to a point that your customers start noticing them," Hayes says. Such visibility results from customer call activity and intelligent network elements; and in Adelphia’s case, from a homegrown analytic cable modem tool called NEMOS. The latest step has been to turn that intelligence into action. What was previously ad hoc activity, Hayes says, has been organized into three areas: upstream performance, i.e. mitigation of ingress, common path distortion (CPD) and impulse noise; realignment of particular nodes that may require either complete sweeps or recertification; and (forthcoming) deployment of Data Over Cable Service Interface Specification (DOCSIS)-based standby power supply technology. Cox adds logic Meanwhile, taking automated network operations to another level, Cox Communications expanded its relationship with Auspice, leveraging their ongoing joint work in Auspice’s TLX MetaScript development environment. (For more, see Communications Technology, January 2004, page 8.) Cox Director of Network Services Debbie Simmons said in a statement that she expects this expansion to enable Cox "to translate our company’s technical expertise, working knowledge and operations processes into programmed logic for running our operations with a new level of efficiency." "Ultimately, we expect this kind of automation to result in greater service reliability, performance and customer satisfaction," she said. Among all the metrics an operator may track, customer calls may provide the ultimate truth test. Cox’s customer service reputation is well known. Adelphia appears be on track with a hoped-for inverse correlation: "As our preventive maintenance activities are going up, as the performance of our network is going up," Hayes says, "we’re seeing an exact opposite trend in the percentage of calls that are coming into our call centers." Jonathan Tombes