AWS Case Study: CAP – POOQ

In the past, CAP provided OTT services, including media processing and distribution, through third-party solutions. However, after experiencing a massive traffic increase during the 12 games of the World Baseball Softball Confederation (WBSC) in 2015, the company looked for a stable and scalable service-deployment plan that could handle such increases during special events and would allow it to develop its services further. In addition, the company needed an environment in which the infrastructure could be managed flexibly and responsively for the continuous improvement of customer service. Hui-yeol Cho, General Manager of CAP, says, “In the past, we used all-in-one services for media processing and Content Delivery Network (CDN). However, we needed a plan to improve services by applying new technologies to POOQ to better accommodate customer requirements and improve customer satisfaction."

Early in 2016, CAP selected Amazon Web Services (AWS) and AWS Elemental. In January 2017, CAP launched mature-media processing and content-delivery network (CDN) services with Amazon CloudFront, AWS Elemental, and Amazon Simple Storage Service (Amazon S3). By June 2017, it had completed migration of all POOQ services, including its OTT service.

To improve the viewers’ experience, POOQ transcodes the High-Definition (HD) and Ultra-High-Definition (UHD) streams at various bit rates based on the customer’s network and the type of devices used. To enable this capability, POOQ uses AWS Elemental Cloud, the platform as a service built on the AWS cloud infrastructure. POOQ can rapidly deploy live video as it automatically provisions and dynamically scales video processing and delivery.

This requires huge computing resources. Using Amazon Elastic Compute Cloud (Amazon EC2) Spot Instances lets CAP perform transcoding at a low cost. In addition, with various management services such as Amazon Aurora, Amazon Relational Database Service (Amazon RDS), Amazon CloudWatch, and AWS Config, it was able to maximize operational efficiency and improve service availability. In particular, with Amazon CloudFront, it could easily and quickly implement various features, such as signed cookies and URLs, blocking policies by country, and origin protection for safer content transmission. It could also apply HTTPS comprehensively. Furthermore, Auto Scaling lets CAP operate servers flexibly, by scaling Amazon EC2 instances up and down automatically and reacting to predictable traffic patterns with the Scheduled Scaling feature. Since its successful migration, CAP has been developing POOQ 3.0 with a new service architecture and features on the AWS platform. The new system will use the serverless components of AWS, such as Amazon API Gateway and AWS Lambda, to reduce the operational burden and focus on developing customer-oriented services.

Hui-yeol Cho says, “One of the biggest benefits of migrating the POOQ service to AWS was the ability to apply technologies and services freely. We developed the media processing service ourselves, letting us actively apply various technologies and develop new services. On-air VOD is one of the most popular services among POOQ users. This service lets users view any program from the beginning at any time, even while the program is being aired. Since migrating to AWS, we have been able to provide stable on-air VOD.”

After migrating video processing and CDN services to AWS, CAP reduced buffering time—one of the key indexes of a streaming service—by 20 percent, improving customer satisfaction. In addition, Amazon CloudFront and AWS Elemental Cloud give CAP the ability to cope with peak loads of live viewing, optimize live event workflows and 24/7 live linear content with maximum resiliency. Today, the company continues to apply and optimize new technologies such as HTTPS, variable bit rate (VBR), adaptive bit rate (ABR), signed cookies, and georestrictions. This active application of technologies improves the user experience, makes resource usage more efficient, and reduces costs.

Hui-yeol Cho says, “As a leading company in the global cloud computing market, AWS is always prepared, with the open source code, sample cases and documentation that engineers want. While developing and configuring a system architecture that would let us introduce Google's open-source Kubernetes system to POOQ 3.0, we found that AWS provided a reference architecture suitable for AWS's characteristics as open documentation. That was very impressive for us. The reference architectures and use cases available on the AWS Cloud, along with the variety and high quality of AWS services, combined with CAP’s DevOps culture to make us twice as effective.”

He added, “One of the most attractive points of AWS is scalability. As POOQ is a broadcasting service, it experiences rapid traffic increases in specific time bands. For the API system, one of the POOQ’s key services, tens of servers are operated at ordinary time bands, but in peak traffic time bands the number of servers is automatically doubled with Auto Scaling and Scheduled Scaling. After those peaks, the number is reduced to the ordinary level. If we served out of an on-premises data center, we would have no choice but to purchase servers for peak traffic, leading to lower cost efficiency due to low server usage in ordinary time bands.”