Introduction
As online users’ appetite for video content increases, digital businesses are becoming more reliant on video streaming. Faced with stiff competition for clicks and views, businesses are working hard to develop the best possible video content. But even the most compelling videos may be abandoned or ignored by users as a result of performance hiccups like slow load time, playback errors, and constant buffering. To ensure the hard work spent generating amazing content isn’t wasted, website and application owners must know the ins and outs of video streaming. This post will provide an overview of video delivery by explaining how video streaming works, how its performance is measured, the extent to which video quality affects user engagement, and how Azion’s Edge Application can improve video delivery.
How does video streaming work?
Unlike downloading, which forces users to save a full video on their hard drive before viewing it, streaming videos are transmitted a few seconds at a time from a media file that is stored remotely. The act of downloading a partial file or the entire file manifests in long delays before the user can view the content, whereas streaming plays almost immediately. Not only does this save space on the user’s device, it also shortens the time users spend waiting to play a video, allowing the rest of the file to transfer while the video is being viewed. To ensure playback if the connection is lost or the network becomes congested, a few seconds of video are loaded ahead of time, which is known as buffering.
Encoding
Because video files are so large, a balance must be struck between speed and reliability to deliver high-quality video with as little playback interruption as possible. After raw video is captured from a recording device, technologies called codecs convert the video into compatible files, then compress the files for storage and delivery and decompress them for viewing, discarding any unnecessary data in the process.
TCP vs UDP
A similar tradeoff is made with TCP or UDP transport protocols, which are used for transporting data packets across networks. Before transferring data, TCP opens a dedicated connection, and ensures all data packets reach their destination in order. Although this takes time, it results in a reliable, consistent experience. Since UDP does not adhere to these restrictions, it is faster than TCP, but may result in lower quality video. Since both have pros and cons for the user experience, video streaming performance is measured with a variety of metrics.
How is video streaming performance measured?
Although video streaming is better for the user experience, it can be incredibly difficult for developers to get right. From the moment a user clicks on a video, all kinds of things can go wrong, from transmission delays to encoding errors. And while the tradeoffs of slow speed vs. high quality video may vary depending on the video’s content and intended viewers, it’s obvious that degraded quality, long wait times, and interruptions in playback are frustrating for any viewer. The Streaming Media Alliance has defined four key metrics for determining the quality of experience (QoE) of streaming video:
- Start Time: The time between when play is initiated and the first frame is rendered
- Video Start Failure: Failure to deliver video within the time cutoff (~10s) from initiation
- Average Media Bitrate: The average bits per second of video transferred
- Rebuffering Ratio: The percentage of time the viewer experiences rebuffering issues (i.e. video not playing smoothly or linearly due to buffer underflow)
Although bitrate may sound similar to a viewer’s Internet connection speed, the actual bitrate of streaming video will be much lower than this theoretical maximum limit, as other factors, such as network conditions and the video’s resolution, might impact bitrate. A higher bitrate is associated with a clearer and crisper picture, better colors, and a better quality video, especially for high resolution and higher motion content. However, it also results in a larger file size, which can lead to rebuffering if the viewer’s connection is not fast enough to process the file.
QoE vs. QoS
QoE can have a huge impact on whether users will watch (or enjoy) a video, but user-experienced quality can be difficult for service providers to control, as it is often impacted by factors like the viewer’s ISP and device performance. As a result, quality of service (QoS) metrics can be used to pinpoint issues with packet loss, availability, and throughput in each step of the delivery chain.
To bridge the gap between QoS and QoE, monitoring is needed to not only achieve a global view of how users are interacting with applications and how individual users experience video service, but pinpoint the root cause or causes of poor performance, through Real-User Monitoring (RUM). With this in mind, companies can use tools like A/B testing to see how the user experience improves with various implementations.
How does video quality affect user engagement?
How a website performs has a huge impact on its bottom line, and video is no exception to this rule. Deloitte’s 2020 Milliseconds Make Millions report noted that after a one-second delay, users lose focus on the task they are performing, and after 10 seconds, they are likely to abandon the task, often without returning.
When it comes to video streaming metrics, slow start time, rebuffering, and changes in bitrate can lead to abandonment. A post from the blog Streaming Media stated that if rebuffering changes by even 1%, sites can see up to 16 minutes less engagement. And changes to bitrate can be even worse. A 2020 study in Communications of the ACM found that videos where bitrate decreased “exhibit the lowest video watching percentage, largest abandonment ratio and highest likelihood to get abandoned within a few seconds after the occurrence of the impairment.”
How can Azion improve video streaming?
Because video files are so large, delivering them from centralized locations like hyperscale cloud providers can result in massive bills for application owners and poor performance for end users. CDNs, which store copies of content close to where end users are located, are a time-tested part of the video delivery journey because they reduce the distance video has to travel. This significantly reduces network latency by not only shortening the trip, but also reducing the number of hops and the opportunity for network failure across various networks.
However, video delivery is not a one-size-fits-all situation. Different end users’ ISPs, browsers, and devices can have a huge impact on QoE, requiring both sophisticated caching capabilities to ensure the right format and bitrate, as well as the ability to monitor and troubleshoot any issues as they arise.
In addition, as Deloitte stated in a 2020 article, “The growing quantity and sophistication of OTT video content means more traffic, more routing, and a greater need for management, optimization, and prediction across the CDNs responsible for delivering speedy and reliable content.” As a result, traditional CDNs may not be a viable solution for the complex video delivery needs of many site and application owners.
That’s why edge computing companies like Azion can be a great choice for video delivery. Like CDNs, edge computing stores content closer to end users to reduce bandwidth and improve delivery speed. Unlike traditional CDNs, Azion’s Edge Platform enables sophisticated computing capabilities at the edge of the network, such as:
- Edge Application: configuring slice settings to efficiently deliver large files and tailor delivery to end users with advanced caching policies
- Edge Functions: easily create new custom features or choose from prebuilt functions such as A/B testing
- Edge Analytics: monitor user-experienced performance with Edge Pulse and gain global insights into application performance with Real-Time Metrics to bridge the gap between QoS and QoE
Furthermore, Azion’s serverless computing model allows our services to scale automatically to accommodate spikes in usage, resulting in pay-as-you-go service that reduces upfront costs, eliminates resource waste, and ensures service availability.
As a result, our clients, such as UniCesumar, one of the largest education groups in Brazil, have successfully leveraged our products to enhance their video delivery. With Azion’s Edge Application, Edge Functions, and Edge Firewall, UniCesumar was able to gain 70% faster delivery of video lectures to their 190,000+ daily platform users, with 100% available service, even as in-person students migrated to distance learning during the pandemic. To learn more about the possibilities of video streaming with Azion, read the full UniCesumar Success Story here.
