Streaming Media Global
Though fraught with challenges, the move to the cloud for DVR, UI-UX, and ad insertion is helping service providers and operators maximize revenue and deliver a better experience to their subscribers.
In search of increased revenue, cost savings, improved service, and customer retention, service providers are migrating key services to the cloud. The number of TV services that are accessed over the network via a service platform or back end for video on demand (VOD), subscription video on demand (SVOD), and other TV Everywhere services such as start-over or catch-up is ramping up fast. The arguments in favor of the cloud are compelling and the momentum unstoppable, although the shift is not always straightforward and should give pause for thought.
Before we take a deeper dive into cloud DVR (cDVR), ad insertion, and the user experience (UX), it may be helpful to define what we mean by “cloud.” As ever, different people mean different things by it.
For example, “cloud” could mean the use of public virtualized IT infrastructure, such as Amazon Web Services (AWS), on which privately licensed software is operated. Or it could mean that the cloud supplies the actual function being offered. In that case, the underlying IT infrastructure may be still be cloud-based or privately owned to create a “private cloud.”
Ian Munford, Akamai's director of product marketing and media solutions, points out that “Cable and IPTV operator deployments are typically happening in private clouds, which limits the value to viewers. OTT providers such as broadcasters use cloud partners like Akamai to deliver DVR-like services and have done so for many years.”
Another important subtlety is the actual location of the equipment inside the cloud, which becomes very important for certain aspects of TV delivery. There’s a general assumption that because something is “in the cloud,” it doesn’t matter where it is actually located, and that it’s probably centralized somewhere to benefit from economies of scale. That’s not true, and without proper consideration, this could be a very expensive mistake.
“From a technical point of view, cloud can be about moving functions out of client devices and into servers,” says Andy Hooper, VP of cloud, solutions and services EMEA at ARRIS. “In that sense you could argue that a telco service provider has been running a telephone network as a ‘cloud’ service for 50 years. In a business context, cloud also implies outsourcing of operations.”
Essentially, we are talking about moving what used to be done in a box in the living room to the network. The most obvious benefit of this is the ability to tap the cost savings of virtualization and rapid response to changing demand.
Cloud DVR
Park Associates' research suggests worldwide cDVR subscriptions will rise to 24 million by 2018. As Ericsson’s Sarah Paris-Mascicki wrote in TV Technology, this reflects a shift in consumer viewing habits and creates “an imperative for providers to build more flexibility into their services.” A recent study released by Technavio expects the global cDVR market to grow at a compound annual growth rate of more than 30 percent over the next 3 years.
By 2020, Ericsson predicts that more than half of viewing will be time-shifted with providers able to reap double to triple the current benefits their DVR services provide, simply by upgrading to the cloud. Other vendors see similar growth. TV applications solutions provider Accedo, for example, expects more than 50 percent of cDVR penetration to have occurred by 2022.
“When it comes to processing key functions of the client box, such as the DVR experience, in a more centralized way and reducing the requirements of the STB [set-top box] there is a clear trend,” says ARRIS' Hooper. “What’s more fragmented is whether this is outsourced to third parties. Some Tier 2 and Tier 3 operators don’t regard TV as core to their business and are more ready to outsource cDVR functions to someone else’s data center. It’s a commercial decision: build versus buy.”
ARRIS says it has a project with a Tier 1 European operator that is building its own cDVR infrastructure, and the company is working with other clients to build the same capabilities in order to sell a white-label service to other operators.
Edgeware customers, including Belgacom in Belgium and Netherlands’ KPN, are using DVR in the cloud quite extensively, according to CMO Richard Brandon (right). “cDVR is usually one of the top three services their viewers use when consuming TV delivered over an IP network—along with Startover TV and VOD,” he says.
The Benefits of cDVR
The value proposition for operators moving to a cloud architecture is well-documented. It includes financial efficiencies such as lower operational expenses in maintaining STBs and reduced overall capital expenses; increased revenue potential by being able to upsell storage services, multiscreen experiences, etc.; and the ability to reduce customer churn by offering advanced services such as parallel recordings, unlimited storage, and improved reliability.
“The cost of supplying and maintaining home DVRs is becoming prohibitive as users demand more and more storage,” says Brandon. “Many of today’s connected devices don’t have sufficient storage for local recording.”
Operators using Edgeware's TV CDN have achieved increased subscriber satisfaction and loyalty from adding catch-up and cDVR. KPN, for example, tripled its TV subscriber base to more than 1.6 million households since launching cDVR in 2011.
“Additional benefits to the viewer include the ability to access stored programs from multiple devices or locations, escaping from the cycle of upgrading out-of-date equipment and simplifying the buying and configuration experience,” says Brandon.
Ericsson reinforces the capital savings a service provider can generate from theoretically fewer service callouts. It calculates that a typical truck roll costs $75 per subscriber and says cost-efficiencies have been the main driver for the deployment of cDVRs by U.S. operator Cablevision.
However, costs don’t simply vanish with a move to the cloud. Service providers face a huge challenge storing unique copies. “Cloud DVR may bring significant potential for video providers to accelerate monetization opportunities, but that potential quickly loses its luster if storage costs for a large-scale service reach $1,215 per subscriber,” says Yuval Fisher, Imagine Communications’ CTO. That alarming figure is tied up with the thorny legal issue of the legal uncertainties surrounding private/public copies and is the single issue thwarting rapid rollout.
Regulatory Challenges
In most markets, it is completely practical to deploy cDVR services. However, the licensing arrangement for private copies in some countries, notably those in North America, is a sticking point. The pivotal issue is whether a unique copy is required for every subscriber, which in the U.S. is deemed as the standard.
“A private copy system requires a unique copy of a program to be saved for every subscriber that requests it, meaning recordings cannot be shared,” says Itai Tomer, head of the cDVR business line at Ericsson. “Each single, unique copy of the program has to be saved for each user, which requires a huge, growing volume of storage and very high recording and playout concurrency, and that can be problematic to sustain.”
“Technically, it is nonsensical to have to create private copies, but it’s a license requirement from some content owners,” says Brandon. “Where that exists, it makes cDVR more expensive and cumbersome to implement.”
In Europe, copyright laws vary according to region. “In Switzerland, for example, the regulatory framework is very well-defined and in accord with single copies, so you find all the major players like Swisscom and Sunrise offering cloud recording,” says Hooper. “In other places, it’s very much a traditional content rights negotiated issue and therefore a commercial decision as to whether the numbers make business sense for an operator to move to the cloud.”
Imagine Communications has done some calculating. “For content owners that want to stick to the exact letter of the ruling, it means, for example, an operator with a million users and 300 hours of recorded content per user will require 120 Petabytes (Pb) of storage,” says Fisher. “If you just store one physical copy of a year’s worth of content for 500 channels—which is a use case from one of our customers—then it means 20Pb of storage. So whether in private or shared copy mode, there are huge storage costs.”
To make matters worse, he says, delivery to multiple devices in multiple ABR formats ups storage requirements fourfold to six-fold, rapidly becoming an even more expensive proposition.
“For a use case where the aggregate bitrate is 15Mbps with three ABR formats and 200 hours of recorded content for every subscriber, and we assume the cost per TB is $300, then you get a cost per sub of $1,215,” Fisher says.
In another example, taking the aggregate bitrate at 10Mbps with two ABR formats and 75 hours storage per user and a TB at $200, the cost is $135 per user. “Even with a very extreme scenario of very low bitrates, you still end up with a cost of $22 per user,” Fisher says. “Multiply any of those scenarios over millions, in the case of any large service provider, and you can see how the costs quickly stack up.”
Imagine Communications advocates just-in-time (JIT) packaging to encrypt and JIT transcoding to minimize storage costs by 50 percent. “Store assets in a single format but deliver in all adaptive bitrate formats for multiple devices,” advises Fisher. “Operators want the ability to shift between interpretations of the private copy ruling, giving them flexibility at the program level.”
Akamai’s Munford reports that many operators still haven’t been able to secure rights for their cDVRs in a way that allows the benefits of the technology to be fully realized. “Until this problem is fixed, the value proposition for viewers will remain confusing and the economic benefits to operators limited,” he says. “OTT services do not have this problem as rights are readily available or, as is often now the case, owned by the OTT provider.”
The legal framework is evolving to open up the market. For instance, the French Digital Bill, approved in July by the French Senate, encourages the deployment of cDVR in that country.
“New and renewed contracts often add the provision for shared copy storage. Over the next few years, it is expected that most, if not all, non-adversely negotiated contracts will allow shared copy,” says Simon Trudelle, senior product marketing manager at Nagra.
User Interface and Experience
Transplanting user experiences to the cloud offers many of the same advantages to operators, notably the ability to change the UX rapidly and at scale, rather than rewriting the UX for every make and model of CPE, and enabling an operator to innovate a discrete user interface (UI) for every subscriber.
User experiences delivered as MPEG or H.264 streams to every STB with full interactivity enable operators to efficiently deploy services “that are equal to—or better than—experiences that run on the box itself,” according to Murali Nemani, CMO at ActiveVideo.
Let’s take a step back and look at where TV came from before we look at where it’s headed. Consumers are used to systems such as Apple TV, where the UX is underpinned by the computing power of multicore CPU and GPU chipsets. By contrast, the TV world in some cases still is restrained by embedded systems that tried to get the most out of low-horsepower chipsets and minimum memory. CE vendors used embedded software developers to build UIs, and the results often were not pretty.
The arrival of multicore chipsets for customer premises equipment (CPE) allowed operators and TV set makers to use designers instead of developers for the UI-UX. As a result, the TV UX has evolved.
“Most TV UIs are now very heavily graphics-focused and what we would call a televisual experience using ‘posters’ or ‘jackets’ and a lot of picture-in-picture,” says Anthony Smith-Chaigneau, senior director of product marketing at Nagra. “A stunning UI-UX that delivers responsive experiences is now expected. The cloud UI is driving us to a position of compromise because the functionality of a native embedded UI-UX cannot be replicated with today’s cloud UI offering.”
Why Cloud for TV UX?
The arguments for a cloud TV UI-UX are similar to those for cDVR and ad insertion; they can be summarized as follows:
- Virtually unlimited back-office CPU power to implement the UI-UX
- Takes advantage of legacy-deployed customer-premises equipment that is technically less capable than modern devices
- Potential for new, less-capable CPE, as all the heavy computing is done in the cloud
- Since applications are run in the cloud, upgrades are avoided in client devices
- Reduced complexity of managing the different models of CPE deployed on the network
- Application download capability such as download of new STB software to any connected device
Nagra’s Smith-Chaigneau points out that many legacy devices simply do not have the physical capacity to offer a slick UI and great UX, so the industry is looking for ways to fix that problem.
“Ironically, TV Everywhere is addressing laptops, smartphones, and tablets that have enormous computing power,” he says. “So with a cloud UI-UX, are we just talking about the issue of ‘incapable’ STB/CPEs in the field?”
Smith-Chaigneau points out that some operators are large enough to support the cost of deploying advanced services and advanced UI-UX by implementing a middleware in the client STB and by supporting all STB hardware models. “They might also look at using cloud services to reduce their total cost of ownership, but it becomes difficult to weigh the real cost of these services, as they have to support millions of consumers,” he says. “Also, they will have to look at the usage of network bandwidth, balancing between unicast and multicast services.
“It may well be that cloud UX is the solution for small and medium operators who want to deploy similar advanced services without having to bear the cost of implementing a middleware in the client STB—or at least be able to support a middleware that provides mainly the video and audio rendering means: no PVR [personal video recorder], no video gateway to home network,” he continues. “Network bandwidth still remains a challenge, but there might be fewer problems as these operators have to serve a smaller number of clients.”
Cloud UX deployment has its share of technology challenges. Nagra summarises these as follows:
- Latency of the remote control: each action of the remote has to be transmitted to the cloud for processing.
- Limited network resources: if network resources are limited, it is difficult to anticipate the network’s actual load. This is the case for live/linear services where each video stream is unicast. Some cloud UX technologies propose unicast for the UI and multicast for the content. The merge of both streams made in the client device requires relatively powerful devices.
- Concurrency of consumer activities: the industry is still learning about the scalability of cloud infrastructure and its availability to support the peaks created by live events.
In addition, Smith-Chaigneau suggests there is a real question about the simplicity of the STB in that both video and audio still need to be decoded “taking into account the numerous compression and transport formats (HD, UHD, Dolby Atmos, etc.) which requires a variety of computing power requirements.”
Vendors such as ActiveVideo espouse the innovation aspect of cloud-based UI-UX. ActiveVideo points to Ziggo's VOD and catch-up services in Netherlands; trend-driven UIs with multiple tiles of live video on single tuner STBs with Liberty in Puerto Rico; and the complete YouTube experience to upward of 500,000 existing STBs at UPC Hungary.
Nagra questions how open providers of video services will be to being “proxied” by a cloud infrastructure. “For example, video services like Netflix and YouTube have their UI implemented in the client device,” says Smith-Chaigneau. “Will they accept that the UI is implemented in the cloud?”
Ad Insertion
Ad insertion is less mature than cloud DVR, although early adopters are beginning to implement the technologies, and many multichannel video programming distributors (MVPDs) are exploring it.
“It’s early days,” says Ericsson’s Tomer. “One thing is clear, though. Operators agree that changing viewing habits combined with OTT video and innovation in cDVR technology have changed the game for advertising.”
The advent of programmatic advertising has begun a transition for sellers and buyers of ad inventory from an offline transaction world to an instant online experience with claimed benefits of better control, choice, and transparency.
“While traditional ad inventory selling and buying mechanisms exist, we anticipate operators will rapidly adopt the cloud as their default ad insertion infrastructure,” says Sanjay Kirimanjeshwar, head of global marketing at Amagi. “This is not restricted to buying ad inventory alone. The entire workflow of selling inventory, buying ad spots, payments, uploading video assets, managing insertions, reporting, and measurement is in the process of becoming integrated. Numerous third-party services and technology providers are plugging in their products and offerings to make this cloud workflow robust.”
He explains that operators are widening access to their ad inventory by partnering with multiple ad exchange platforms. “Since the systems are cloud-based, it eliminates geographical limitations related to sourcing and delivery,” he says. “For example, media planners based in the U.S. can create, manage, and monitor tailored advertising campaigns for audiences in Canada and Central and Latin America. As the broadcast feeds permeate geographical boundaries, subject to necessary regulatory clearances, operators are rapidly expanding their audience base and reach.”
Operators in the OTT space quite naturally see the cloud as the technology choice for ad insertion, whether for live or VOD services. It’s worth noting Nagra’s observation that with the exception of the U.S. market—where MVPDs actively manage some of the advertising space on behalf of broadcast networks—“Demand for addressable ad insertion remains low as the ad space is managed by broadcasters,” according to Nagra’s Trudelle.
Cloud deployment (meaning ad insertion software deployed on virtual machines) can be used for both broadcast and multiscreen ad insertion. This is happening in three different ways:
- Server-side ad insertion for live streaming, cloud DVR, time-shifted, and VOD services
- Client-side ad insertion for pre-ingested ads onto the origin or on the cloud
- Instances of integrations with digital video ad networks (such as Google or SpotX that traditionally serve video to the web) with video service provider networks
“Client-side” ad insertion relies on the client requesting an ad to be streamed to the device at each ad break. “Server-side” ad insertion inserts the ad in the viewing stream as part of the actual program. One advantage of service-side ad insertion is that it is far less likely that the viewer can use ad-blocking software to override a client, and the advertiser can be sure its ad was actually delivered.
The key advantage overall is the ability for operators to create more value for advertisers by enabling delivery of targeted and personalized ads.
“Advancements in server-side ad insertion, especially for live sports and news content with abrupt ad breaks, are catching operator attention,” says Kirimanjeshwar. “Likewise, operators are beginning to serve personalized ads on VOD platforms where subscriber profiles are predetermined. Either way, operators can offer an enhanced experience to both advertisers and viewers.”
Scalability is another key advantage. Cloud simplifies the addition of new ad exchanges and integration with demand-side platforms, and it supports ad insertions for a growing audience base compared with traditional and offline models. For some, the major benefit is wresting full control and visibility over ad insertions.
“The other development among operators is the adoption of programmatic spot ads,” says Kirimanjeshwar. “Compared to the earlier 1-minute spot ad inventory model where ad sourcing was largely localized, the introduction of cloud technologies has allowed aggregation of spot ads. Now, operators can sell spot ads programmatically.”
Since the storage is cheaper and computational speeds are higher, cloud-based systems should be able to process audience information faster and deliver targeted ads accordingly, in a more cost-effective fashion for operators.
There are challenges, though. Like cDVR and timeshift services, inserting different ads for different viewers creates personalized and unicast streams, potentially unique to each viewer. Streaming ads from a centralized point can put strain on the delivery network because viewers generate their own streams of traffic right across the network. Additionally, it’s important that the ad is sent in the same format and bitrate as the program it is being inserted into.
For cDVR, Brandon explains that Edgeware’s customers are looking to solve this problem by distributing ad-insertion functions closer to the edge of their TV service infrastructure. “The ad decision-making is hosted in the cloud, but the ads themselves are stored locally and inserted at the edge of the network in real time,” he says. “Where, of course, they are not vulnerable to client-based ad-blocking software.”
Perhaps the greater impediment—a speed bump rather than a roadblock—is on the business side and the attempt to align various stakeholders (broadcasters, brand advertisers, and measurement firms). The feeling is that the value of TV Everywhere is directly aligned to the ability to manage addressable ads and that as this becomes more transparent, the whole industry will shift.
This article originally ran in the Autumn 2016 issue of Streaming Media European Edition as “Electrifying OTT Services With the Cloud.”
