More than a pose
Online yoga destination Glo uses AWS to distribute its classes worldwide
Glo is a subscription-based yoga studio in Santa Monica, California and offers an online video library of more than 5000 classes. Topics include yoga, meditation and pilates, and the courses are taught by renowned instructors from around the world. The online library is accessible for streaming, downloading and viewing from the Glo website and via iOS or Android apps from £14 ($18) per month.
The studio’s on-demand video assets range from individual classes to full four-week programmes and require scaling video processing from tens to hundreds of hours of content a month. Ahead of a recent corporate rebrand, Glo recognised the need to re-encode and re-edit each of its video courses to bring all on-demand content currently with the brand up to date. Every video in the Glo library now includes newly branded, four-to-six second opening bumpers.
Given the scope and size of the Glo video archive and the need to scale video processing over time, the company initially considered outsourcing its video encoding needs. Ultimately, Glo decided to overhaul its entire video production pipeline.
The studio reached out to AWS premier consulting partner and managed service provider, Onica, which introduced the Glo team to the suite of AWS Elemental media services. Glo leaned heavily on cloud services from AWS to streamline its content production processes and support its growing video library.
“We looked at all of the options on the table, but ultimately decided that the project provided the perfect opportunity to retool our entire infrastructure, and AWS played a huge role in that,” says Nery Orellana, Glo’s cloud architect. “There’s no way we could have programmatically re-encoded our library in such a quick time frame without AWS Elemental Media Services.”
“When we want to add on to our service, we can easily make changes. we don’t have to rebuild the workflow from scratch”
To retrofit existing video assets with new bumpers and branding, Glo crafted a workflow in which the team spliced content into an L cut (where the audio from one scene continues playing over the next), edited out the old bumper and spliced in the new branding using a custom-built framework. The process makes it easy to dynamically re-encode content as further changes are required. The video encoding and distribution workflow then began with content residing in Amazon Simple Storage Service (Amazon S3).
Once content was located to Amazon S3, it ran through an AWS Step Functions serverless orchestration service for approval and was transcoded using AWS Elemental MediaConvert, a file-based video transcoding service with broadcast-grade features, which allows Glo to easily and reliably transcode on-demand content for multiscreen delivery at scale. The service can function independently or as part of AWS Elemental Media Services’ family of cloud services.
In order for Glo to more intelligently service geographically dispersed customers, Amazon Route 53, a scalable Domain Name System (DNS) service, performs geospatial lookup to identify the regions from which members are trying to access content. This allows Glo to provide a better localised server by storing content in regionally relevant S3 buckets.
“Having AWS power our video workflow has helped us dramatically streamline and operationalise costs,” adds Jeff Neil, Glo director of engineering. “By using reserved transcoding slots, we can scale to accommodate large volumes of content. Looking at a per-minute cost analysis, it’s much more cost efficient for us to use AWS Elemental MediaConvert versus other tier-based solutions, which tend to result in unnecessary overage charges.”
AWS Elemental MediaConvert incorporates quality-defined variable bit rate (QVBR) encoding – another draw for Glo. Since most of the studio’s classes are also made available for offline viewing, the company saw an advantage in using QVBR for video processing.
“When you’re accessing a piece of content and don’t have a great internet connection, great deliverability is a must. QVBR helps us dramatically reduce the payload size of offline content to mobile apps. We can easily shrink the file sizes so that when users are streaming the classes, they’re getting a consistent, high-quality viewing experience without using any unnecessary bandwidth,” explains Neil.
As Glo’s business grows, the company aims to augment its offering with new capabilities, from livestreaming and captioning to support for popular streaming devices. It fully expects AWS tools to play a big role in accommodating these demands, contributing towards making the Glo video workflow even more intelligent.
“AWS Elemental MediaConvert and the AWS platform have given us greater flexibility and building blocks for future opportunities. When we want to add on to our service, we can easily make changes, so we don’t have to rebuild the entire workflow from scratch,” concludes Orellana.
This article originally appeared in the July 2019 issue of FEED magazine.
