Performance Enhancement of a Global Audio Streaming Platform

The customer is a leading global audio streaming platform with millions of active users. The platform provides a wide range of services, from live radio stations and music streaming to podcasts, news coverage, and sports games broadcasting. The platform is available across 200 connected devices, including web, mobile, voice assistants (Amazon Alexa, Google Assistant, and Microsoft Cortana), home sound systems (Bose and Sonos), smart TV, gaming consoles, and smartwatches. The client was looking to improve the service quality of their platform, make it more user-friendly on all supported devices, and enable it to handle high volumes of requests simultaneously.

The client, a leading global audio streaming platform with millions of active users, was facing challenges in improving the service quality of their platform. The platform, which provides a wide range of services including live radio stations, music streaming, podcasts, news coverage, and sports games broadcasting, needed to be more user-friendly across all supported connected devices such as Amazon Alexa, Google Assistant, etc. The client also wanted the platform to handle high volumes of requests simultaneously, especially during peak times. The platform was available across 200 connected devices, including web, mobile, voice assistants, home sound systems, smart TV, gaming consoles, and smartwatches. The growing customer base necessitated an improvement in platform usability and efficiency in handling high amounts of requests.

The solution involved migrating the on-premise data center to the cloud and refactoring the infrastructure code. The client's platform code was refactored from .NET Framework to .NET Core. The on-premise data center was migrated to AWS by moving all workloads to Kubernetes while maintaining several EKS clusters and supporting ECS clusters. Backup solutions were developed using Amazon S3 in the multi-account environment. The infrastructure that had been written for Windows was migrated to Linux to reduce licensing costs. A multi-level cache was utilized to allow the platform to process up to 20,000 requests per second. Scaling up during peak hours was achieved with the use of HPA and different scaling types. Amazon CloudFront was used to cache the content in edge locations to ensure no delay in content delivery. Alexa Skill Lambda was re-implemented with new dialogs and action choices and Amazon Alexa Skill Account Linking was performed to unify the subscription approach.

• The migration of the on-premise data center to the cloud and the refactoring of the infrastructure code resulted in a significant reduction in operational costs. The platform's performance was boosted, with the ability to handle high volumes of requests at peak times without any delay in content delivery. The usability of the platform across different connected devices was improved by adding new features like Alexa Skill Account Linking, which unified the subscription approach. This allowed users to link their Alexa user identity with their identity in other services (in this case, the client’s platform) and pay only for one subscription instead of two. The overall user experience was significantly improved.

• Reduced operational costs by migrating the on-premise data center to the cloud and moving platform infrastructure from Windows to Linux

• Enhanced platform performance by enabling it to handle up to 20,000 requests per second

• Improved user experience by implementing Alexa Skill Account Linking that unified the subscription approach