Managing content metadata and media for a TV service is an increasingly complex business: more and more content sources are aggregated, fragmented metadata needs to be unified and enriched, media assets have to be transcoded for many different end-user devices, UXs and recommendation engines are more sophisticated and information-hungry than ever.
In order to help TV operators and TV-related companies cope with content operations, Optiva Media has created MediaSuite, a one-stop solution for media management which consists of four interrelated products:
- MediaStream: Integrated management of content metadata and programming information
- MediaTag: Automatic content tagging and categorization using AI
- MediaShot: Unassisted generation of content and people images from media files
- MediaCode: Transcoding workflow management in the cloud
As of 2019, Spanish operators Vodafone Spain and Orange Spain are already using one or more MediaSuite components as part of their TV operations workchain. Typical traffic figures of a MediaSuite deployment are:
- 200+ daily updated TV stations
- ~10,000 active VoD content entries from catch-up and external catalogs (Netflix, HBO, Amazon Prime, etc.)
- Central metadata repository with an accumulated 0.5M content and 150K people references and 1M+ associated images (covers, banners, actor pictures, etc.)
- More than 1,500 hours/month of transcoding work for internal VoD asset processing
In order to keep up with these growing flows of metadata and media processing, MediaSuite needs to rely on an architecture with maximum scalability and reliability and a high degree of subsystem modularity.
Why Amazon Web Services
MediaSuite was designed from the ground up as a native AWS solution. Two strategic factors convinced us AWS was fit for the job:
- Dependability, measured as a combination of scalability, cost-effectiveness and reliability. AWS orientation towards serverless computation with AWS Lambda and SQS ranks very high along these three dimensions. Classical redundancy, uptime and security KPIs in critical areas such as RDS are also unparalleled in the market.
- Innovativeness: Some key components of MediaSuite make extensive use of AI technology for textual and video analysis. Beyond core cloud functionality, AWS has stayed on the edge of value-added service offering with a constant track record of new launches such as Elasticsearch integration, Rekognition and AWS Elemental MediaConvert. This commitment to innovation makes AWS an ideal choice for the long-term evolution of MediaSuite.
We mention some of the most important AWS capabilities used by MediaSuite products:
- Mediastream’s central metadata repository (known as the IMS), is deployed on top of Amazon Aurora. Import (from metadata sources to the IMS) and export (from IMS to end-user platforms) subsystems are heavily parallelized via extensive use of AWS Lambda Functions pipelined through SQS queues: we are now refactoring the remaining server-based elements (running on EC2) to achieve a fully serverless architecture. Metadata matching, a process by which new content entries are compared against the IMS for duplication detection and information merging, uses algorithmic machinery reliant on Elasticsearch. The online UI is served through Amazon API Gateway.
- MediaShot relies on Amazon Rekognition for detection and extraction of relevant snapshots from media files to be used as graphical metadata (covers, banners, etc.)
Media processing management by MediaCode is designed around the workflow capabilites of AWS Step Functions. Heavy-duty work, such as transcoding, quality checking and packaging, is also executed by EC2-virtualized services or transferred to AWS native services like AWS Elemental MediaConvert. Intermediate and deliverable assets are stored with S3 and archived in S3 Glacier.
Against the tradition of TV operators to rely on on-prem infrastructures, Optiva Media’s MediaSuite has proved our customers that fully cloud-hosted operation solutions derive significant advantages in terms of reliability, scalability and cost effectiveness. Thinking long term, serverless architectures guarantee an optimum degree of adaptability and modularity, while the host of value-added services offered by AWS, such as Rekognition, paves the way for ambitious product evolution based on AI analysis of metadata and media assets.