The demands of video delivery are pushing content caching further and further toward the edge of the network. And I can't help but wonder if the edge will eventually reach as far as our own living rooms.
Netflix Inc. (Nasdaq: NFLX) has pushed the envelope with content delivery networking for years now, combining direct connects at interconnection sites with caching appliances that sit all the way inside the last-mile network of an Internet service provider. Recently, the streaming company also detailed some of the ways it's optimizing those delivery methods including the use of proactive caching, which helps Netflix preload content for viewers when network traffic is low.
Today, Netflix organizes appliance caches into clusters and preloads a master appliance in each cluster with content based on what it believes will be heavily streamed, like a new original series. Then those master appliances fill the content caches of peers in their clusters so that each cache doesn't have to access a content source farther away. Peer-to-peer transfers currently take place across Internet exchange points and between embedded appliances in ISP networks. But what if, in the future, the tail end of those clusters extended into consumer homes? That would allow not only for content to be pre-positioned within a home, but also for homes to seed other Netflix streaming devices in the same neighborhood.
The idea is particularly interesting given the fact that Netflix is purportedly preparing to offer a download-to-go feature in the coming months. Caching is just a fancy way of describing content that's downloaded and stored for later use. (See Netflix Queues Up Video Downloads.)
There are problems with my theory. The first is that media streamers remain cheap, in part because many don't include on-board storage. And particularly as higher-resolution video becomes more popular, storage requirements for content will increase. (Though of course storage costs are always declining.)
Second, the idea of using algorithms to determine what's likely to be popular among a large group of people is not the same as determining what should be preloaded into an individual home. Caching content for a large group means preloading is presumably worthwhile as long as a certain percentage of users view that video. However, caching at the individual level means that if a user doesn't view the content, or maybe only a few neighborhood peers do, then preloading becomes a less efficient use of resources.
Third, and related to the second point, establishing a controlled peer-to-peer network among user-owned devices may be more trouble than it's worth. Managing a network of appliances in a data center is one thing. Managing a network of retail devices sitting in user homes is something else.
Let's look at the potential advantages, however. Proactive caching at home could mean even less network congestion at peak times, and Netflix could form clusters in places like apartment buildings and other multi-dwelling units where the probability of a nearby cache hit goes up.
Plus, proactive caching at home would improve consumer performance. When you want new content, it would already be there waiting.
To be clear, there's no indication that I've seen that Netflix is planning in-home proactive caching. Even if the cost/benefit analysis worked out, Netflix undoubtedly has plenty of other priorities.
But, given that Netflix is always tinkering with network delivery, it also wouldn't surprise me to see some type of trial partnership with a device company, perhaps in a specific geographic region. Maybe with TiVo Inc. (Nasdaq: TIVO) in a cable partner's territory?
Don't forget, Netflix is also bringing its service to the Comcast X1 platform. Comcast Corp. (Nasdaq: CMCSA, CMCSK) hasn't wanted Netflix appliances placed within its last-mile network, but that's because it would have no control over those boxes. If Netflix content was instead proactively cached on an X1 set-top with Comcast's input over the process, maybe the calculation would change. (See Comcast Confirms Netflix Coming to X1.)
At the very least, it's an intriguing idea.
— Mari Silbey, Senior Editor, Cable/Video, Light Reading