Helping Streaming Video Steam Ahead

There's a reason, other than free WiFi, that people appreciate having coffee shops on virtually every corner. A local source for a dose of caffeine limits waiting. For the coffee shop owner, less waiting means happier customers -- who will return.

Demand for OTT content in the US and abroad is gathering more steam than an espresso machine. Juniper Research Ltd. forecasts that subscriptions from OTT video providers, such as Netflix Inc. (Nasdaq: NFLX), will generate $31.6 billion globally by 2019, up from just under $8 billion in 2014. And Reed Hastings, CEO of Netflix, recently predicted the demise of traditional TV sometime between 2020 and 2030. At the same time, research firm MTM reckons that Netflix's US OTT market share will drop to 50% by 2018 as new streaming contenders emerge, fragmenting the market further.

We are in the midst of a fast-paced industry transformation where the traditional Internet/last-mile divide is blurring as many historical roles change. This transformation is being powered by the Internet streaming software stack on the one hand and an ever-growing global CDN footprint on the other.

However, the move to ubiquitous Internet-based broadcasting is still fraught with delivery problems that have yet to be solved. Too many network operators -- and content publishers -- are underestimating the challenges associated with delivering bandwidth-hungry Internet applications (both content and services) across the last mile with high enough quality to meet rising consumer expectations. And today's issues will only get bigger as the volume continues to grow and mobile becomes an even more prevalent mode of viewing.

What are those challenges and considerations? They include:

  • How to efficiently and cost-effectively deliver HD and UHD/4K video over the Internet on a scale comparable to current pay-TV consumption, while assuring a buffer-free experience

  • How to reduce end-to-end cost per bit of content delivery in the gigabit world, where the top 15-20% of subscribers consume over a terabit of data per month

  • How to manage and deliver unicast-based live streaming over the Internet -- both regular linear programming and major events like the NFL's Super Bowl or the upcoming 2016 Summer Olympics -- on a scale of tens of millions of households

  • How can operators migrate from end-to-end "managed video" to tomorrow’s "open garden" with subscribers consuming both on-net and Internet-based services, while maintaining high quality and potentially offering a unified user interface

  • How to reconcile a distributed content delivery architecture with the centralization that some SDN/NFV approaches bring

These are big issues. How can a network operator best address them? This is where a timely, customer-focused approach -- a local approach -- comes into play. Even in a gigabit world, last-mile speed alone is not enough. It's the quality and performance of the apps that matter most. Like all those coffee shops, the shorter distance that streaming video and other OTT applications need to travel, the better consumer experiences will be. And of course, consumer experience is a huge retention factor, which is so important in today's competitive environment.

Adding virtualized local content delivery capabilities to broadband access networks puts popular Internet-based information and entertainment on a faster "expresso" lane to consumers. Local content delivery brings OTT content and services to the edge of operator networks, closer to consumers, thus shortening latency and improving quality and application performance. It also relieves network congestion, conserves bandwidth and lowers operating and capital expenses for mobile, cable, telecom and other broadband service providers.

Augmenting broadband access networks with local content delivery helps carriers optimize their bandwidth use and decongest the last-mile grind. From an operating standpoint, local content delivery must integrate easily with both fixed and mobile networks, and be available as a virtualized elastic solution for maximum efficiency. It is also a key enabler for last-mile network providers to profit from the variety of business models that will abound -- those that exist today and those that are yet to come.

Even beyond their first cup of morning "Joe," people hate to wait. People also have high expectations for the products and services that matter most to them, including streaming video. Local content delivery is the approach that operators should take to meet those expectations.

— Alan Arolovitch, CTO, PeerApp

kq4ym 12/11/2015 | 10:44:20 AM
Re: Long tail Assuming the prediction by Netflix of "the demise of traditional TV sometime between 2020 and 2030," one might rightly wonder just what will be the scenario of entertainment and knowledge consumption in five to ten years. Will be be using portable devices or even larger screens, and what might be the mix. Will compression algorithms get such improvement that those gigabit connections might not be necessary for most applications?
MikeP688 12/5/2015 | 7:20:10 PM
Re: Long tail thanks for your kind words @aarolovitch024 and your updated stats.    I am not sure about your views on Netflix (starting to be adapted) as we're seeing others making major inroads. I would agree with your contention that the trend is towards streaming--although it still has to evolve.


kjsing 12/5/2015 | 2:26:07 PM
Re: A 3-punch solution perhaps Hi Aarolovitch024,

Analyses show that the combination of just in time transcoding and just in time packaging does not only result in significant gains in storage efficiency, 70+%, but also in bandwidth efficiency, 40+%. The question of which is cheaper, CPU cycles or storage bits, is ligitimate in case of SVoD. For streaming of life programming any storage is very transient, i.e., buffering, and this question is not an issue.

As for adjusting bit rates dynamically, this is done to match instantanously available network bandwith to help smooth out congestion. In a live streaming setup you don't want to multicast multiple copies of the content at different bit rates, but adjust bit rates and/or resolutions only as close as possible at the subscriber.
aarolovitch024 12/5/2015 | 11:24:49 AM
Re: Long tail Wholeheartedly agree. The big question for me is how the OTT ecosystem is going to evolve and manage the fragmentation that is coming - in terms of aggregation, content discovery, billing.
aarolovitch024 12/5/2015 | 11:22:17 AM
Re: A 3-punch solution perhaps Kjsing, just-in-time transcoding comes to solve ABR storage fragmentation issues. I am not convinced that edge storage efficiency is a major issue, nor that peak-time CPU is cheaper than storage. Lastly, how decreasing bitrate is good?
aarolovitch024 12/5/2015 | 11:12:46 AM
Re: Long tail Good for you! According to Conviva Viewer Experience Report, percentage of video streams that buffered went up from 26.9% to 28.8% between 2013 and 2014. Full two thirds of all streams failed to achieve highest available resolution. Streaming is in the early phase of adoption compared to PayTV - only 10-15% broadband subs stream on a sustained basis, the live events are in low millions of concurrent streams, direct-to-consumer services outside of Netflix started to pick up only now - and the end-to-end delivery infrastructure doesn't hold sustained 1080p yet, never mind 4K.
kjsing 11/30/2015 | 10:25:31 PM
A 3-punch solution perhaps Potentially the solution can be found in an SDN/FNV switch/router model that does three things simultaneously:

1. Prioritizes isochronous data over non-isochrouns data.

2. Instantly transcodes video where congestion is anticipated or occurs.

3. Packages the video at the edge in the streaming protocol that the client device prefers (HLS, MPEG-Dash, etc.).

Prioritization of isochronous data acts as a traffic cop as well as a load balancer for transcoding resources. In addition it also acts to keep latency within bounds.

Instant transcoding reduces the bitrate of the incoming video. It is reasonable to assume that the trajectory from the origin server to the edge will have a decreasing bandwidth profile. Instant transcoding eliminates transmission of unncessary copies at different bit rates from origin server to CDN and eliminates storage of these copies in the CDN.

Packaging the video at the edge similarly reduces transmission bandwidth and storage as instant transcoding does. Although MPEG-Dash is supposed to be the standard protocol, it is reasonable to postulate that HLS and other protocols will continue to be in use for a while.
MikeP688 11/29/2015 | 3:26:02 AM
Re: Long tail As I am writing to you now, I am "consuming" what you've envisioned.  The quality is fabumous, the uptime is 100% and I have had no issues whatsoever on it.    It has to be sustainable--no question.    
danielcawrey 11/27/2015 | 4:30:59 PM
Long tail There is enormous opportunity in the "long tail" market for OTT. This is similar to what Amazon did with the book industry.

There will be OTT offerings for every demographic. One of the most lucrative will probably be in sports. There are so many sports fanatics out there that will pay for these services. And they will be incredibly loyal.