OCTASOFT

Case study

High-Definition Media Streaming Infrastructure

A leading media company needed scalable infrastructure for high-definition video and audio streaming, integrated with proprietary on-site hardware.

Industry
Media & Entertainment
Duration
18 months
Technologies
High-definition streaming path and the point where network throttling was foundProprietary on-premises streaming hardware sits at one end of a sustained high-bandwidth link into an AWS VPC, where EKS runs the streaming workloads. Alongside it, a Fargate harness generates sustained test traffic against those workloads. Prometheus collects metrics across the path and CloudWatch collects AWS-side metrics. The point on the link between the VPC and the on-premises hardware is marked as the place where network throttling appeared under sustained load, and a line from Prometheus to that point shows how it was observed. Terraform modules, kept DRY, define the environments beneath.On-premisesProprietarystreaminghardwareAWS VPCEKSStreaming workloadsFargateLoad-generation harnessSustained test trafficPrometheusMetrics across the pathCloudWatchAWS-side metricsNetwork throttlingfound under loadTerraformModules kept DRY, so the test environment matches production
A simplified view of the streaming path and where the throttling surfaced. Illustrative, not as-built documentation.

The challenge

Streaming quality is unforgiving: problems that would be invisible in a typical web application show up immediately as dropped frames and audible artefacts.

The difficulty was not throughput in the abstract. It was sustained throughput, from cloud to on-premises equipment, without the kind of intermittent degradation that is very hard to reproduce.

The approach

We designed and implemented an AWS architecture with networking optimised for sustained high-bandwidth transfer, then set about proving it under realistic conditions instead of synthetic ones.

Load testing ran on Fargate, which made it cheap to generate properly heavy traffic on demand. That testing surfaced the problem that mattered: AWS network throttling that only appeared under sustained load, well past the point most test suites stop. Until then it had presented as unexplained quality drops. Prometheus instrumentation made it visible and attributable.

Infrastructure was defined in Terraform throughout, with modules structured to stay DRY across environments so the test environment closely resembled production.

The outcome

  • Network performance bottlenecks affecting video quality were eliminated
  • Cloud and on-premises systems integrated without bespoke glue that only one person understood
  • Infrastructure provisioning time fell by roughly 80% through automation
  • Monitoring became good enough to diagnose the next problem, not just this one

The durable result was diagnostic capability. The throttling was found because the instrumentation was there to find it, and that instrumentation outlasted the engagement.

Next step

Facing a problem like this one?

We work on platform engagements where the constraints are real and the outcome is measurable. Describe yours and we will tell you whether we are the right people for it.

Book a consultation