Using Capella Systems’ Cambria FTC and Cambria’s floating licence mechanism, Adstream manages transcoder demand with a tiered method that scales vertically and horizontally.
Simple transcode queue scaling
Cambria FTC’s efficient transcoding queue sizes and its ability to prioritise transcoding per individual job are the first step to dealing with peaks and managing bottlenecks. Using the Cambria FTC API, workflow engines can set the number of concurrent transcode jobs to match the specifications of any situation, creating the right balance of throughput to avoid needless I/O bottlenecks. By coupling queue optimization with per-job priority settings, Adstream ensures that JIT and SLA critical content can be transcoded ahead of lower-priority content.
Vertical scaling
Using the Cambria floating licence API, a licence can be attached to a job from a pool of available licences. Then, as a new job replaces a completed one, that licence is transferred to this new instance as the previous instance is terminated. As the default Cambria licence model supports instances up to 64 cores, Adstream can continue to increase the instance size appropriately and without disruption to the BAU. When demand reduces, the instance size can be decreased appropriately, or when no longer required, terminated completely, releasing the licence back into the pool.
Horizontal scaling
While it is perfectly possible for content held in S3 storage in any of the AWS regions to be transcoded in any of the Adstream AWS regions, this cross-regional processing is not practical at scale. Adstream clients’ regional peaks coincide with the path of the sun. A pool of floating licences available to all regions ensures that spikes in demand can be handled within the region to allow load balancing and further optimisation of the transcode queues. Adstream’s automated deployment mechanisms protect against any misconfiguration or variances between regions. All the tools found in one AWS region need to be available in all other regions to provide a consistent product as part of DR strategies. Capella has helped greatly in this respect by automating update installs and providing the ability to set features via an API. This reduces the complexity of evaluating new software releases and features in UAT environments before they’re deployed into live production.
Workflow simplification
Using Cambria FTC’s evolving toolset, Adstream has been able to reduce the number of subsystems required to meet the specifications and requirements for each broadcaster or playout provider. Audio mapping and linear audio normalisation can be quite complex but are necessary steps, as are dynamically creating slates and providing custom head builds. Using the Cambria FTC API, Adstream does all this and transcode in a single step. Not only has this consolidation optimized speed throughout the system, the lean and powerful infrastructure reduces the number of points where issues can arise, making support easier to manage. Cambria FTC’s ability to transcode content directly from and to S3 buckets has also resulted in further efficiencies, especially around handling longform content and provisioning large amounts of temporary storage to stage source and target files.