As executive director of advanced access architectures for Cox Communications Inc. , Finkelstein is constantly looking at how to maximize investments in the capex-heavy access portion of the network, and he encouraged the audience here to consider "regrettable" and "non-regrettable" investments -- the former being something that only solves a customer problem short-term and will soon be replaced.
"As we go through the planning, what are the things we are going to do that we are not going to regret doing or if we are going to regret it, make it a choice," he urged, and not merely a financial expediency. High on Finkelstein's "non-regrettable" list is pulling as much fiber as possible, potentially to pedestals very near the customers. That isn't always the natural instinct of access network planners and builders, however.
The current rollout of DOCSIS 3.1 on the existing cable plant will enable operators to offer gigabit services, to keep up with their telecom competitors, but as take rates for those services increase, there will be constraints on available upstream bandwidth -- something that is becoming more critical in an era when folks commonly create their own content. Finkelstein calls the upstream bandwidth limitations the "iceberg" for DOCSIS 3.1 -- which he insisted will last much longer than the Titanic.
There are interim steps to addressing those issues -- moving from 1K-QAM to 4K-QAM, for example -- but cable operators will reach a point where they need to split their 500-home nodes into smaller segments and push fiber deeper into their networks, he said.
"As you are pulling this fiber, this is the opportunity to build the network of the future -- and our natural tendency, particularly of the outside plant folks, is to just overlash yourself and pull it down and when you need to, you'll pull it off the strand and connect it into a tap," Finkelstein said.
Instead, pulling a fiber through a pedestal and pulling enough to support a long-term strategy of fiber-to-the-home makes more sense, he added. "One of the things you need to give serious consideration to when you do these splits, do you have a PON strategy or an FTTH strategy, -- EPON, GPON, Ethernet, wavelengths, whatever you want it to be?"
There is a lot of latent capacity in the coax plant, Finkelstein acknowledged, but cable operators need to be realistic about the customer demands that lie ahead. Historical data shows a 49% compound annual growth rate for bandwidth and that would mean being able to deliver 2 Gbit/s per service group by 2021 and 10 Gbit/s by 2026.
"You won't regret having that bandwidth" because it will produce a better customer experience, he said.
Finkelstein also cautioned against delaying deployment of DOCSIS 3.1, for financial reasons, saying such delays will inevitably accelerate the need for node splits. Instead, he advised starting with a "micro-segment" strategy that targets investment to areas where there are identified problems, based on customer complaints and truck rolls, then moving to capacity deployment by year two and ubiquitous use by year three.
"Coax has a long life and a long useful life but how do we optimize each and every spend we have an prepare ourselves for the next step?" he said. He pointed to Cox's decision to deploy switched digital video and a one gigahertz plant. "Those were spends we never ended up regretting. Some of it [was] because of some very smart folks -- we are standing on the shoulders of giants in the cable industry but now it's our turn to prepare it for the next step for the people that will be standing on our shoulders to keep it going and have all the capacity our customers will need."
— Carol Wilson, Editor-at-Large, Light Reading