Dolby Hybrik’s encoding architecture on AWS is designed as a Platform-as-a-Service (PaaS). Unlike many other cloud encoders that require you to upload files to their servers, Hybrik launches its media processing engine directly inside your own AWS account.
The platform primarily utilizes the following AWS services and components for its encoding workflows:
1. Compute: Amazon EC2 (On-Demand and Spot)
The core of Hybrik’s encoding power comes from Amazon EC2 instances.
- Instance Types: It typically uses compute-optimized instances (like the C5 or C6i families) because media transcoding is a CPU-intensive task.
- Spot Instances: Hybrik is famous for its “Spot Market” integration. It can automatically request AWS Spot Instances—which are up to 90% cheaper than On-Demand—and includes built-in failover logic. If a Spot instance is reclaimed by AWS, Hybrik automatically moves the task to a new machine.
- Segmented Encoding: Hybrik can split a single large file into many small “chunks” and process them in parallel across dozens or even hundreds of EC2 instances simultaneously, drastically reducing the total time to encode.
2. Storage: Amazon S3
Hybrik uses Amazon S3 as both the source and the destination for media.
- Because the processing happens in your VPC, there are no “egress” fees for moving data between your storage and the encoder.
- It supports S3 Event Notifications (via SNS or Lambda) to trigger “Watch Folder” workflows, where a new file upload automatically starts a transcoding job.
3. Orchestration & Security: IAM and VPC
- VPC (Virtual Private Cloud): The EC2 encoding nodes run within your private network, ensuring that your mezzanine (master) files never leave your security perimeter.
- IAM (Identity and Access Management): You grant Hybrik’s control plane permission to “spin up” and “terminate” EC2 instances in your account using IAM roles.
4. Communication: Amazon SNS
Hybrik uses Amazon SNS (Simple Notification Service) to send alerts and status updates. When a job is finished, failed, or reaches a specific milestone, Hybrik can push a JSON notification to your own backend or email.
Comparison: Hybrik vs. AWS Elemental MediaConvert
It is important to distinguish between the two, as they are often used in the same breath:
| Feature | Dolby Hybrik | AWS Elemental MediaConvert |
| Model | PaaS (Runs in your account) | SaaS (Managed by AWS) |
| Pricing | Flat monthly fee + your EC2 costs | Per-minute of video processed |
| Customization | Massive (Complete control over x264/x265) | High, but limited to AWS-supported features |
Why Hybrik doesn’t use Lambda for Encoding
Encoding is a “long-running” and “resource-heavy” process, which clashes with the core limitations of Lambda:
- Execution Limits: Lambda has a hard timeout of 15 minutes. High-quality video encoding (especially 4K or Dolby Vision) often takes much longer.
- Compute Power: Encoding requires sustained, high-performance CPU usage. Hybrik uses specialized EC2 instances (like the C5/C6i family) because they offer better price-to-performance for constant heavy math than the serverless Lambda environment.
- Cost at Scale: For processing petabytes of video, EC2 Spot Instances are significantly cheaper than the equivalent compute time in Lambda.
How Lambda is typically used with Hybrik
While the Hybrik engine itself runs on EC2, engineers often use Lambda to “glue” the system together:
- Triggering Jobs (Watch Folders): An S3 “Object Created” event triggers a Lambda function. The Lambda then takes the file metadata and sends a POST request to the Hybrik API to start a transcoding job.
- Preprocessing: Lambda can be used to check if a file is valid (e.g., checking file size or extension) before paying to spin up a Hybrik EC2 instance.
- Post-Processing: Once Hybrik finishes a job, it sends a notification via Amazon SNS. A Lambda function can subscribe to that SNS topic to move the finished file to a CDN or update a database (like DynamoDB).
Summary of Component Roles
| Task | Service Used |
| Heavy Transcoding | Amazon EC2 (managed by Hybrik) |
| Orchestration/Glue | AWS Lambda (written by you) |
| Job Notifications | Amazon SNS |
| Asset Storage | Amazon S3 |