By Feed Staff | Feed Magazine | Published Issue 30 / August 2020 Edition
Pixel Power divulges the practicalities, advantages and disadvantages of cloud playout in the latest edition of FEED Magazine.
The broadcast industry has a habit of lusting over the shiniest, newest thing and jumping for it without taking the necessary steps to get there beforehand. Right now, that shiny new thing is cloud playout, a flexible and scaleable solution for broadcasters wanting to spin up channels faster, deliver traditional and OTT content more efficiently, and ensure a low-cost, reliable disaster recovery strategy (just in case there’s a pandemic or something).
But is it a destination all broadcasters need to go to? If so, how do we get there and what do we need to consider? Pixel Power, a Rohde & Schwarz company, answers these questions and delivers a comprehensive journey to cloud playout, with comments from CEO James Gilbert.
Broadcast technology has evolved to the point where it’s possible to deliver linear and non-linear content from the public cloud. Up until quite recently, a linear playout chain was built from separate pieces of dedicated hardware and controlled using serial or Ethernet cabling. Broadcasters could choose the different components from various manufacturers – a ‘best of breed’ approach – and would expect their chosen automation system to control them and make them work seamlessly together to deliver their required output. As general-purpose computing power increased and became more affordable, it became possible to implement certain functions of a hardware-built playout chain within software; creating ‘channel in a box’playout devices, which were capable of collapsing several hardware functions into a single device. However, graphics, DVE and subtitling would still rely on dedicated hardware boxes, as these functions are more specialised and require extra computing power that goes beyond what an underlying general-purpose computer can provide.
Today, general-purpose computing power is continuing to increase and IP standards for video and audio streams are evolving fast. New integrated playout vendors have emerged with experience in areas like graphics and subtitling, and hardware devices have been completely replaced by software modules, with SDI transports replaced by IP streams. The dedicated hardware is gone, and the complete linear playout chain is pure software. This means that it can be virtualised and deployed in a private data centre or public cloud.
DATA CENTRE VERSUS PUBLIC CLOUD
“Cloud means different things to different people. For some, it’s a private data centre, made up of racks full of standard computer servers; for others, it’s a public cloud provider like AWS, Azure or Google, which make their own computers,” explains Pixel Power CEO James Gilbert. A private data centre invariably offers broadcasters more flexibility in infrastructure than a public cloud provider, since it may be possible to install dedicated hardware equipment or even some SDI infrastructure if needed. Public cloud implementation is different: there is no possibility of changing the infrastructure, so the solution must be able to work within the constraints of the specific cloud offering.
But there are many reasons broadcasters are considering deploying some or all of their content delivery in the public cloud. Gilbert says: “If you’re a media business, your focus isn’t going to be on running data centres, it’s going to be on creating content; content that inspires your viewers. You don’t want to be concerned with building and maintaining a big infrastructure project.” The public cloud can be purchased on a usage basis, so there’s no need to invest up front in hardware infrastructure; it can be scaled up or down depending on changing market and business needs; and most public cloud vendors offer a range of different virtual machine types with varying degrees of computing power if more features need to be added. Furthermore, most public cloud vendors offer a wide range of locations, which may be useful for legal, commercial and practical purposes.
If there is a festival or sports event that a broadcaster wants to cover for a short amount of time, they can quickly and cost-effectively launch a new channel in the cloud and take it off the service line-up once the event is finished. In addition, cloud playout simplifies disaster recovery, since a broadcaster can contract a cloud back-up more cost-efficiently than an on-premises playout. This is because a cloud backup has the option to start up and create the operational cost only in the event of a disaster, whereas an on-premises backup needs a dedicated pre-investment for a disaster event. There are some disadvantages – the biggest one being cost. Certainly, for a 24/7 broadcaster, cloud deployment would cost more over a three-year period than hosting an equivalent infrastructure on their own premises. Gilbert explains: “If you’re running a 24/7 operation, you don’t need the kind of flexibility that a public cloud provider would give you to start up a channel, run it for a few hours and then shut it down. It makes far more sense to invest in your own infrastructure, because it will be cheaper than renting it by the hour or reserving it for a year.”
When implementing a cloud-based playout, there are considerations that need to be made. For example, there are many ways to use public cloud infrastructure for your playout. You can rent it and be responsible for the software installing and operation yourself; or you can purchase a fully managed service through a third-party vendor who purchases the infrastructure from the cloud provider. Gilbert enthuses: “For optimum cost reduction and greatest flexibility and control, running your own cloud-based infrastructure is preferred, provided you have the skills to build and manage it.” Deploying to and operating cloud infrastructure requires specific new skills that may not be available in the organisation – although it might be possible to upskill existing staff or seek help from the solution provider to assist with the migration process.
Gilbert says: “You need new skills; it’s a whole new language. It’s important to understand the architecture of the provider and how best to use what it has got to offer, as well as its limitations – because there are things you can’t do in the cloud that you can do on-premises. For example, multicast network is a big no-no in the public cloud. Although, it’s more of a deliberate limitation to prevent a rogue virtual machine engine affecting other customer’s virtual machines in the cloud. It provides isolation and protection, and it’s good practice to have even within a broadcast facility, but it does cause some constraints for those that require that functionality.”
Another consideration is security, since you’ll be putting valuable content and probably consumer information covered by GDPR in the hands of an external company, rather than keeping it in-house on your own servers. But the reality is that the leading cloud providers have very robust security and encourage their customers to use it. “The security team at even a small cloud provider have a skill set and experience far greater than that of a major national broadcaster and there should be comfort in that,” insists Gilbert.
Then there are technical considerations, such as deciding the data centre location. There may be legal or regulatory constraints, such as latency and the availability (or lack of) high-bandwidth connectivity. Choosing the right IP standards for your cloud playout is also a challenge – the standardisation process often lags behind vendor technology offerings, leading to the risk of adopting something that later becomes nonstandard or obsolete.
“A lot of standardisation effort is around ST-2110 uncompressed video, but this is generally not a viable option for cloud playout; cloud network infrastructure is unlikely to support the required bandwidth and packet throughput and the egress cost of the uncompressed video would be prohibitive. A more likely mezzanine format of 30-50Mb H264 RTP MPEG-TS is well-established and a better fit for cloud infrastructure,” says Gilbert.
Monitoring and control also need careful consideration. The automation clients could be cloud hosted, with simple remote desktop access from any location to interact with them; or they could be installed locally, with appropriate connectivity through a firewall to the cloud playout infrastructure. Gilbert says: “If you distribute via satellite or terrestrial, you need to bring your feed back down to the uplink point, so that might be more suitable for local monitoring and control. For internet distribution, it probably makes more sense to keep monitoring and control fully cloud-based.” It’s also unlikely that one provider will give you everything you need, and in any case, it might be wise to choose other parts of the overall solution from alternative vendors. Integration might be required with a scheduling system, a MAM, a transcode farm or a CDN/VOD platform. The ability to draw upon significant computing power at will could support AI add-ons, such as automatically adding metadata to video and audio content in a media library. But whatever the integration, it’s important that the chosen supplier is able to support licensing and implementation on the cloud platform.
Finally, there’s the question of availability. When you have your own equipment in your own rack room, your availability is clear. The picture is not so clear with cloudbased infrastructure and you will need to ask your provider important questions, like does it have sufficient capability to give you what you need to run the channels? Availability might not be there unless you pay for it, perhaps by reserving critical parts of your infrastructure to guarantee availability. In a disaster recovery situation, it’s possible that other local broadcasters or businesses might use the same data centre, so if a natural disaster happens, the data centre spare capacity is likely to be used up quickly.
Moving playout to the cloud is perfectly possible and there are clear and straightforward ways to make it happen. However, despite the end goals being the same, every broadcaster’s processes are different, so careful planning is key. The shiny new thing is shiny for a reason – it is worth getting there, but for the right benefits. Our industry did it before, moving from tape to file-based workflows, and reaped the rewards. Getting the perfect playout solution for you relies on working with trusted partners along that path.