From Capex to Opex in Broadcast Distribution How engineering decision-makers balance the priorities of reducing capex and speeding up time to market while maintaining control of quality. date
Page 2 of 11 Famine and Feast These days, the broadcast business has a big, fat target on its back. For viewers, it is the Golden Age of television, with greater choice and higher quality than ever before. For the companies that create and distribute content, the market is changing at a speed and intensity that makes it hard to keep pace. Viewing content on demand anytime, anywhere, on any device is the choice for most viewers, and audience fragmentation is accelerating. Viewers still consume a lot of TV, and advertisers must reach an audience at scale. But viewers have become agnostic about the channels they watch, the devices on which they watch and when they tune in. The Pew Research Center reported that 78% of Americans watch local TV news and 73% get their news from networks or cable news channels. Alternatively, 61% check news online, and 54% from radio or local newspapers. In October 2017, Nielsen reported that total viewership for the National Football League was down nearly 8% over the first six weeks of the season compared with the same period last season. Commentators have suggested that the decrease is tied to the controversy over players kneeling during the American National Anthem but the league also lost 6% of its regular-season viewers from 2013 to 2014. 1 Fifteen million viewers, on average, are still tuning into NFL games, making them a product of immense value, but if conventional audiences are shrinking for the NFL, what franchise is safe? 2 Fundamental Question Content production costs are never spiraling down, and the demands of distribution are increasing every month. In response, broadcast and cable are adapting to digital disruption in the same way as every other long-tenured business. They are searching for greater efficiencies in process and improved productivity in distribution. That has led them to focus on finding ways to convert high capital expenses, which are traditional to broadcast and cable distribution, into predictable operating expenses that scale with volume and revenue. Moreover, they need to do it without sacrificing
Page 3 of 11 control over the technical quality of their content from the time it leaves the broadcast center to the time it reaches the viewer. For engineering decision-makers, it comes down to a fundamental question: How much control do I need over every link in the distribution chain? The traditional answer was given by the head of distribution for CNN nearly two decades ago, when he told a conference audience, If I could control every single piece of it, I would. If I could afford to own the satellites, I would. But I can t. As distribution has undergone fundamental changes from analog to digital, digital to IP, IP to ipad and to the cloud the decision-maker s answer has also evolved. Financial executives to whom they report no longer understand why contribution and distribution should travel over pristinely engineered dedicated networks. It s all data, isn t it? Why aren t we using the same big data pipes that the IT department buys? That question has a good answer, but the trend is clear and the pressure to produce better outcomes at lower cost is not going away. The Video Headend If disruption is inevitable, broadcast decision-makers want to ensure that they oversee it and that they continually strike the right balance as technology and corporate priorities evolve. That is leading them to rethink the video headend. The video headend is the main core of video management, and it controls a variety of functionalities, including ingest, storage, management, and packaging. The video distribution infrastructure carries compressed video in HD or SD formats, encoded with different compression algorithms, and bundled with various types of audio and
Page 4 of 11 additional data services. Operators must normalize this diverse content for delivery to the last-mile network. Once the video signal is captured, it needs to be ingested; in the case of linear content, the different channel streams are received and decoded, and in the case of nonlinear contributions, video assets are transcoded. During the reception of video signals, video and audio contributions are monitored to control the quality of the ingest. In case of alarms, a mechanism promptly requests a channel or network diagnosis. Prior to delivery, content-dedicated hardware performs the encoding and encryption to ensure the content is protected during distribution. Multiplexing functionalities allow efficient resource utilization by aggregating many individual content streams into one. Software, Not Hardware Traditionally, this complex workflow required an equally complex array of specialized devices, daisy-chained together and controlled with a mix of analog and digital systems. Costs were high, management was difficult and hardware failure was an everpresent concern. Recently, however, software virtualization has made possible a revolution in video headend design. The revolution was a long time coming. The company that rebirthed IBM s idea of virtualization, VMware, was founded in 1998 and went public in 2007. Virtualization creates a "virtualization layer" that can run across multiple computers and operating systems, treating them as a single computing resource. It makes it possible to transfer processes once accomplished in hardware into a 100% software environment running on servers, and to take advantage of software's much greater flexibility and faster innovation rate. Applying software solutions to the encoding, multiplexing and distribution of content makes it possible for the broadcast industry to grow its content delivery capacity without adding to costs for channel processing hardware and data center operations. This architecture offers the speed and efficiency to deliver broadcast content in real time, near real time and on-demand. Software virtualization is a big first step toward reigning in capital expenditures, improving processing efficiencies, and enabling process automation. Software is no panacea, however, nor is the cloud the answer to every storage and distribution requirement. Content producers rightfully insist on tight control over the creation, storage, processing and distribution of the content that provides their revenue stream. They need to integrate software in virtual environments that can hybridize existing systems and connect with satellite and fiber networks as extensions of the cloud, offering a robust and reliable path to growth in capacity with minimal cost.
Page 5 of 11 Benefits of Virtualizing the Headend A virtual system consolidates the video processing for all services into a single platform. That simplifies the operational structure, creates a service-level view across all operations, and reduces the management of what used to be technology silos. Lower and more efficient capex The complexity of the traditional video headend was acceptable in a time of slow change. Video processing technologies from multiple vendors were each dedicated to single services, with an overlay of additional management systems to provide visibility and troubleshooting. As distribution volumes and methods multiply, capital expenses (CAPEX) spiraled upward. A virtual processing environment renders the multiple devices and management tools as software managed by a unified interface. The unified interface allows for virtual content processing to be configured as microservices, and these microservices can be added, configured, changed and taken down with mouse clicks. The need for physical equipment does not disappear, but more and more of it may be implemented as commodity computing resources that are available off the shelf. Workflow rationale also improves operations, and realizing the workflow with virtually implemented micro services contributes to lower operating expenses.
Page 6 of 11 Time to market A virtualized headend is less a collection of devices than a pool of computing resources that can scale dynamically based on demand. That is a good fit for a market where content providers need to be able to introduce new programs, services and features in short order to meet the expectations of consumers accustomed to frequent updates and fresh content. New services or channels can be deployed in minutes, and the installation and hardware interconnection issues vanish. That creates time to test, refine and then roll out the workflow needed to originate and distribute the content. Agility, quality of service and new revenue opportunities Virtualizing the headend, whether in onsite hardware or the cloud, makes it easy for operators to establish a set of channels on demand. The headend interface captures the key data video source, desired output formats, quality levels and scheduling and then provisions the resources, selects the right processing platforms, and requests the required network connectivity. Reliability is also improved because logical network elements can dynamically handle failures without service interruption. A malfunction can be spotted as it occurs and resources immediately redistributed to deal with it. These capabilities open new revenue opportunities. It becomes simple to create and dynamically change licensing options for content, from a one-time fee to subscription, and even adding a pay-per-feature capability for accessing premium content within a standard subscription. Content streaming to a phone and to an HDTV could be priced differently and managed effectively in the virtual environment. The Vector Virtualized Headend Globecomm has introduced a virtual headend that can be installed at a broadcaster's own facility or provided as a managed service on the secure Globecomm cloud, which also interconnects as a Tier 1 partner with Amazon Web Services. It was a strategic design decision meant to solve the dilemma that engineering decision-makers face when virtualizing their operation. If the only goal is to drastically cut capex, 100% virtualization in a public cloud can deliver the goods. But if the goal is also to maintain the right amount of control over access to the content and end-toend quality of service, keeping critical parts of the process in-house becomes a compelling proposition. With Globecomm s Vector hybrid design, content owners get to determine the capex/opex balance that makes the most sense. Designed to sharply reduce the capex requirements for a fully in-house installation, Vector also integrates with the
Page 7 of 11 Globecomm cloud and, through it, has Tier 1 access to AWS services and content distribution capabilities. Because processes are virtual micro services, content and processing elements can be configured to run anywhere in the environment, and reconfigured as content owners gain experience with virtualized processing. What is Vector? Globecomm s Vector is a virtualized video headend that simplifies the processing, packaging, and customization of video content for multi-platform delivery across DTH, OTT, IPTV, cable and terrestrial TV. The virtual headend architecture provides the ability to scale the system to deliver hundreds of content channels, and can be constructed using commercial off-the-shelf (COTS) servers, or blade server components occupying significantly less rack space and data center management requirements than traditional encoding, multiplexing and channel processing hardware.
Page 8 of 11 Vector virtualizes encoding, statistical multiplexing and service provisioning. These are all processes that require a significant amount of rack space in a traditional content processing data center. The nature of the virtualization process increases the capacities that any single server can handle by dividing the compute (CPU, Memory, I/O) resources amongst the installed virtual instances (software applications, including encoder, mux, compression) and reducing the number of physical servers needed to perform the required tasks. As a system, Vector provides the ability to reliably handle the processing needed to output SD, HD (1080), and High Efficiency Video Codec (HEVC) encoded video. The system s flexibility allows for combinations of these encoded channels to be simultaneously processed and output to the distribution channels. The distributed content may be satellite content channels that have been downlinked, transcoded and turned around for additional affiliate delivery. The system may also be used to receive content via fiber networks outputting processed content for cable TV distribution or handed off to a content packager for delivery via an OTT channel. The system also allows for the processing of channel content acquired by satellite or fiber and delivered to an existing terrestrial over-the-air system. The goal of implementing the Vector system is to drive channel distribution diversity and capacity without adding significant amounts of hardware on a per channel basis. Installation in an Existing Data Center Installing Vector in a current data center significantly reduces the amount of physical hardware that is required to encode, and mux the output delivery channels. The Vector system does depend on server capacity to run the virtual server instances that provide encoding, compression, and mux functions; however, the hardware configuration is driven by the channel mix and features that need to be provided (e.g., loudness control, logos, etc.). A hierarchy of computing power is required based on the content formats that are delivered, and those content types present different resource loads on the computing equipment. The lowest resource load is processing SD channels. The SD channel output is delivered as a variable bit rate (VBR) stream with a chroma subsampling of 4:2:0 streaming at 1.1 Mbps. The next tier would be to process and output HD channels. These channels would be output using an H.264 (MPEG4-10) encoded stream that is delivered VBR with a chroma subsampling of 4:2:0 streaming at 4.2Mbps. The baseline server configuration using these channel video parameters can output 80 SD channels and 36 HD channels for distribution.
Page 9 of 11 The baseline server configuration capacity would consist of four single RU servers for a single thread, non-redundant system. The addition of redundancy would provide two more servers to the configuration, with any extra IP switch capacity that may be required for connectivity. The core configuration provides the virtual infrastructure to deliver up to 150 channels in combinations of SD and HD to the existing data center. The base system may easily be configured to exist within half of a standard 7-foot equipment rack. Globecomm offers engineering support to provision and the materials necessary to grow this system to meet more channels, as required. As a product in your datacenter, Vector is engineered, configured and tested to meet the content delivery requirements of your network and delivery capacities. Globecomm provides the installation and personnel training necessary to operate the system, and is always ready to assist with upgrades and support. As a value-added feature, Globecomm offers 365x24 monitoring and Network Operations Center (NOC) support for customers that elect to have this level of assistance. Provisioning a Managed Service Globecomm provides both turnkey system integration and managed service products for our customers. The Vector platform can be delivered to meet content delivery requirements as an on-premise owner operated system, or as a service that is provided as a cloud hosted system. As a cloud-based software as a service (SaaS), Globecomm hosts Vector in our datacenter. The advantage of this is that it provides direct connectivity to both satellite and fiber networking, enabling service enhancements for content reception, satellitebased retransmissions and content origination, as well as a connection point to Century Link (formerly Level3), a Tier 1 Content Delivery Network (CDN), for internet content distribution. Globecomm will engineer the hosted system to meet your delivery requirements and provide personnel training on the system operation. Additionally, Globecomm has partnered and connected with Amazon Web Services for hosting and SaaS offerings. Through this partnership, Globecomm has the capacity to leverage the web service resources of Amazon to build, host, and deliver content to worldwide regions and availability zones. Globecomm provides on-staff personnel that are Amazon architecture certified to ensure customers build the most costeffective solutions using the Amazon suite of tools and services. Amazon CDN is a valid option for content distribution from the hosted Vector system.
Page 10 of 11 Future-Proofing the Broadcast Center There s a popular phrase in the information technology world: data is the new oil. Engineering decisionmakers in media and entertainment are certainly aware: All the big IT vendors and consultants are offering ways to abandon the old approach, using software and putting it all out on the cloud. Overall spending on public cloud services is forecast to reach $160 billion in 2018, up 23% from the previous year. 3 More than most businesses, however, television is about the experience delivered to viewers minute by minute, hour after hour. The room for error is minimal, and the stakes in terms of advertising spend are enormous, despite the ongoing fragmentation of the viewing audience. The cloud cannot do it all. Acquiring, processing and distributing content will take the combined resources of onsite data centers, cloud storage and processing, dedicated fiber and public content delivery networks, along with the world s satellite infrastructure. With the introduction of Vector s unique system and service flexibilities, virtualized processing power and diverse multi-network distribution capabilities, Globecomm is equipping content owners and multichannel video programmers with a complete solution to drill for that new oil to reap the benefits in terms of lower capex, greater speed and flexibility without surrendering the essential control over content at every stage that television demands. About Globecomm Globecomm is a trusted global connectivity partner for designing, managing and distributing voice, video and data solutions to the most remote locations on Earth under the most treacherous conditions. The company s multi-network satellite, fiber and cellular infrastructure is the backbone of mission-critical RF and IP communications for government, maritime, media, enterprise and oil & gas customers in over 100 countries. In addition to operating managed network and hosted switch services, Globecomm designs and integrates best-of-breed broadcast and OTT media solutions; complete enterprise communications and data management systems, including Internet of Things applications; and on-premise and cloud-based enterprise video platforms. We proudly serve U.S. government and
Page 11 of 11 NGOs around the world with Morale, Welfare and Recreation (MWR) services, and assist with enterprise and government disaster relief efforts. Headquartered in Hauppauge, NY, Globecomm has locations in Texas, Florida, Maryland, New Jersey, Virginia, the Netherlands, South Africa, Germany, Singapore, Hong Kong, the United Arab Emirates, Indonesia and Afghanistan. More at or GSIHQSales@globecomm.com. About the Author James Brown, Product Manager - Media Platforms, plays a key role gathering customer requirements and defining how the media services and products of Globecomm can exceed the expectations of customers, in terms of technology innovation and cost competitiveness. He has previously designed satellite telecommunications equipment, end-to-end communication systems, and software systems. He has been a contributing member of the Globecomm team since 2013. 1 "NFL Losing Viewers at Alarming Rate but Faces Limits on Its Response," by Brian Goff, Forbes SportMoney, October 23, 2017. 2 "NFL Ratings Continue Fall, Latest Nielsen Numbers Show," by Darren Rovel, ESPN, October 18, 2017. 3 Public Cloud Services Spending Hits $160 Billion This Year IDC, by Ugo Onwuaso, CommunicationsWeek, January 19, 2018.