This blog is the second in a three-part series on achieving operational scale for your streaming business. Read part one here.
Whatever the use case — live event or VOD, ads or subscriptions, one viewer or millions — the key to a successful streaming service is choosing reliable technology with built-in redundancy that automatically scales operations and seamlessly delivers high-quality streams. Our last blog reviewed the key components that any streaming service would need to scale operations (and compete) in the industry. These essential elements — all part of the Uplynk platform — include a plugin architecture full of rich APIs for automation, dynamic personalization (such as DRM, ads, or localization), channel creation and distribution support for all endpoints (including FAST platforms), and a seasoned team that can provide support or services to help you ramp up fast and scale down just as quickly to conserve resources. In this blog, we want to dig deeper into the playback scale.
Before diving in, let’s quickly review what makes streaming unique. The best comparison is traditional TV: a mature industry with thorough guidelines and standards. Feeds are sent to satellites or headends and delivered over dedicated networks, making cable or satellite TV a consistent, high-quality experience. On the other hand, streaming content is sent over the public internet and requires the video to be sliced into smaller chunks for delivery. When a viewer hits play, a manifest file is delivered to the player, containing instructions on retrieving the content chunks. To put into perspective how much work is required to stream content, for a 30-minute video, a manifest file would include 450 (4-second) or 900 (2-second) chunks (for low latency goals). There’s also a significant difference between live vs. VOD content. With VOD, the content is ingested, sliced, encoded, and stored ahead of time. Because of this, once a viewer hits play, a manifest is generated and delivered from beginning to end, including ad breaks. But for a live linear or live event feed, content is ingested, sliced, encoded, and stored in real-time. Only a partial manifest is generated for the beginning video chunks when a viewer presses play. The manifest will continue to update (i.e., new chunks are added while older, already consumed chunks are dropped) until the content ends.
Any video content provider wanting to broadcast (or multicast) their content to a group of people can efficiently distribute it using one manifest file with compatible protocols like RTMP. Still, it comes with tradeoffs such as packet loss, poor viewing experiences (QoE), and limited ability to apply DRM, insert targeted ads (which generate higher CPMs), localize content, or troubleshoot effectively. Additionally, it almost always requires additional network configuration, making unicast (i.e., one-to-one or OTT) easier and the most common form to stream content over IP. However, this approach also comes with challenges, like adding more nuanced workflows to reach more viewers, devices, and platforms.
For example, protecting your content against piracy can complicate workflows by requiring multiple formats to support large audience requirements. Older players that don’t support DRM (Digital Rights Management) need HLS and AES-128 encryption. Older iOS devices require HLS and FairPlay. Newer iOS devices support HLS and FairPlay, and CMAF CBC. Older Windows and Android only support CMAF CTR. Newer Android, Windows, and iOS should support all CMAF formats. Your content must be packaged in multiple formats to allow playback on all devices. DRM is one of many factors that can disrupt your workflows, requiring you to put time, money, and resources toward building, maintaining, and improving a solution. Your infrastructure should allow you to streamline your workflows to deliver consistent experiences across all devices and platforms at scale.
While two viewers might be similar, they are never the same. Even though they may be tuning into the same channel, they’re likely in different locations, use different devices, experience different network conditions, and have different interests and viewing behaviors. Sending your video content as unicast means you have the potential to tailor each experience by generating a unique manifest for each viewer with content, ads, recommendations, and precise DRM or blackout management, all based on your viewers’ device type, location, history, demographics, and other data, such as configured business rules or instructions.
Milliseconds before a chunk plays, data is used to decide on the most optimal stream to deliver, from the best audio or video variants to seamlessly replacing content due to a regional blackout. If we return to our earlier example, a 90-minute live sports event with 2-second chunks contains approximately 2,700 chunks in the manifest – for every viewer. Multiply that by 10 million concurrent viewers, and that’s 27B chunks that need to be tailored and delivered flawlessly. This requires a technology provider that can manage infinite business rules and reliably execute multiple real-time decisions for every viewer, so there’s never an interruption in service.
We architected our platform from the beginning to individually manage every viewing experience at scale. It’s like an automated, built-in concierge service for all your viewers. Our performance relies on our manifest servers to generate a unique playlist of content, ads, and playback instructions for each viewer. Based on the user agent, IP address, instructions received by each client device, and the data passed via simple URL parameters, and our technology makes thousands of distinct decisions, including which ad configuration to use, whether or not to apply replacement rules such as blackouts, and which digital rights management configuration should be enforced. We designed our technology for the industry’s most challenging use cases, including reliably delivering and managing hundreds of thousands of high-profile live events at scale for over a decade.
Read our technical article for a deeper understanding of how Uplynk is built to handle audience surges or support low latency goals, robust enough to manage blackout restrictions and content rights on a per-viewer basis, and powerful to support personalization capabilities such as content recommendations, ad breaks, and other localizations like closed captions.
Uplynk is a highly-scalable and reliable streaming platform with solutions and services that can integrate into any workflow so you can create powerfully differentiated OTT experiences – all with a fraction of the resources. We offer a flexible range of workflows, from simple one-click configurations to more advanced APIs that trigger workflows for notifications, job processing, and additional integrations like machine learning. Uplynk is the foundation for broadcasters and streaming services to evolve quickly, extend, and scale their offerings.
Contact us today to learn more.