Dec 18 2019
Streaming services dominated headlines in 2019 with the launches of Apple TV+, NBC ‘Peacock’ and Disney Plus as well as the increasing investments in Netflix, Hulu and Amazon Prime developing original content. With more people in the US subscribing to a streaming service than a traditional TV subscription, it has become evident that the market for streaming content will boom for the foreseeable future. While this is good news for the tech and entertainment companies, it poses a hurdle for the content delivery networks (CDNs) that support them.
CDNs have relied heavily on a traditional cloud model for support, but with the significant growth they’re experiencing, they must invest in new infrastructure to deliver. Edge computing decentralizes infrastructure and puts compute and storage resources in smaller deployments closer to customer endpoints, minimizing data movement, reducing latency and downtime. However, edge computing does not solve all issues and requires data movement. As edge platforms rely on the traditional Von Neumann computing architecture, there is an inherent size limitation that makes it impossible to add the needed additional CPU processing.
To deliver on the stringent demands of CDNs, edge platforms must adopt a new architecture. Computational Storage is a new innovative solution to today’s architecture, in which compute moves closer to where data is generated, rather than the data being moved up to compute. This approach reduces data movement in edge environments employed by CDNs and speeds up the time to analyze petabytes of metadata for high capacity-driven, read-intensive applications such as streaming content.
A typical CDN traffic flow involves lots of data movement and processing spread out over a variety of edge infrastructure. Steps include reading the metadata, confirming its location and resolution, encrypting the data and re-writing the metadata back to storage. This CDN traffic flow is simplified with Computational Storage. The Computational Storage drive reads the metadata, verifies multiple users simultaneously and encrypts the data and stores the encrypted data for transmission. With In-Situ Processing, this is all done within the storage device itself.
Computational Storage provides CDNs with the opportunity to do much more than just smoothly supporting streaming video. Imagine a service that delivers consumers an interactive, three-dimensional display, including virtual reality. To achieve such a feat would require significant boosts in capacity and processing power for both CDN edge infrastructure and endpoint devices. Computational Storage can provide those improvements with compact form factors and no excess hardware requirements.
CDNs need to be able to rapidly provide a variety of encrypted and access-controlled content to subscribers. Computational Storage will become a fundamental solution to the architecture. Learn more about this technology from other use cases here.