Reliability through simplicity
Cable operators have long been stigmatized by a perception that their networks are inherently unreliable, yet, reliability has always been a major industry concern. Some of the very first SCTE conferences in the '70s focused on reliability because, at that time, it was common to have 40- or 50-amplifier cascades that, frankly, weren't all that reliable. Failure of a single component in an amplifier or power supply could interrupt service for thousands of downstream customers in a community.
To add to the challenge, the industry was attempting to wire the nation and yet was not considered credit-worthy. As a result, a huge amount of capital investment had to be financed out of cash flow from operations. For engineers, this translated into: "Use it up, wear it out, make it do." Lapses in service quality, while irritating to customers and franchising authorities, were considered an inevitable part of the business. Nevertheless, engineers struggled to find ways to improve reliability—supertrunks and CARS-band microwave were often employed to shorten amplifier cascades.
The '80s brought the franchise wars, with each applicant trying to out-promise its competitors. Proposals often featured microwave and standby power supplies, in addition to ever-escalating channel capacities. Some vendors developed fully redundant amplifiers, with castings the size of streamer trunks. These units were remarkably expensive and, curiously, remarkably unreliable. The complexity required for their sensing and switching circuitry produced many additional failure points. The standby power supplies of that era often introduced their own outages through component failure, or worked only until their batteries ran down during long utility outages, without alerting the operator that they were in standby mode.
The advent of hybrid fiber/coax architecture, with its passive fiber links to nodes serving very short coaxial amplifier cascades, has led to a dramatic improvement in inherent cable system reliability. It has also given us an opportunity to rethink our approach to reliability. Experience indicates that the fiber portion of this system is exceedingly reliable. Laser and optical receiver equipment, at the headend, hub and node, has very few components and rarely fails. The fibers themselves are housed in a tough package that almost never fails, unless it is forcibly severed. While very rare, these cuts are quite disruptive, and many operators have reduced this risk by installing redundant fiber rings to hubs serving tens of thousands of homes, and, in some cases, a redundant fiber path to every node. While this gives rise to the question about whether the redundant fiber switching electronics is more prone to failure than the likelihood of the fiber itself being cut, experience indicates that fiber cuts account for a very small reduction in service availability.
That leads us to focus on the reliability of the coaxial plant itself, and its connection to the power utility. As noted above, standby supplies themselves have often been the cause of cable system outages. While the surviving manufacturers in this area are building units with improved reliability, there are still many components involved, and the likelihood of an equipment failure in a standby power supply is measurably greater than that of a very simple ferro-resonant transformer. Added to this is the fact that we no longer need to protect long downstream cascades with standby power supplies. There is a significant probability that a utility outage will also knock out power to the TVs and PCs in our customers' homes, making continued cable service moot.
The exception comes when we provide lifeline telephone service. If we do indeed offer that service, and our marketing brethren feel it's imperative that we provide plant powering or local battery powering for phone service during utility outages, then we have little choice but to provide standby power supplies. This adds significant cost to the provision of telephony service, particularly in the plant powering case.
Amplifier reliability in an HFC plant goes to the heart of the matter. Early in the evolution to HFC, major amplifier vendors tried to sell the same complex modular amplifiers that they had designed for tree-and-branch cable systems. New entrants began building very simple, non-modular amplifiers. With only two- or three-amplifier configurations, virtually every need in HFC plant could be accommodated. Because these amplifiers had comparatively few components, and no connectors for modules or diplex filters, they turned out to have significantly lower mean time between failure than traditional amps, and were much cheaper.
There's a lesson here. Rather than boot-and suspenders redundancy, a very high level of reliability can be obtained by keeping all elements of our outdoor transmission plant extremely simple, with minimal parts count. It is Time Warner's experience in Rochester, N.Y., where we are operating lifeline telephone service, that this approach can yield in excess of 99.99 percent service availability.
There are two fundamental approaches to reliability: Redundancy and minimizing complexity and parts count. While there are places where the former must be embraced, the latter can be highly effective and provides a huge bonus: dramatically reduced costs. The touchstone of our design philosophy should be "reliability through simplicity."