Architecture
The same platform in three places. Where it sits decides what it costs to run, and the decision is reversible.
Three deployments
What runs where, in each of the three.
| Tier | On-premises | Hybrid | Cloud |
|---|---|---|---|
| Apps and APIs | Your servers | AWS | AWS |
| Encoding | Your GPU | Your GPU | AWS |
| Storage | Yours | Yours | AWS |
| Delivery | Your edge | Your edge | CloudFront |
| Content keys | Your building | Your building | Your building |
Which one fits
Most operators start hybrid, and the parts that cost money stay with them.
On-premises
Lowest running cost. Nothing leaves your network, and there is no charge per gigabyte.
- Apps and APIsYour servers
- EncodingYour GPU
- StorageYours
- DeliveryYour edge
- Content keysYour building
Hybrid
Cloud absorbs a sudden crowd. Storage, encoding and delivery stay on your hardware.
- Apps and APIsAWS
- EncodingYour GPU
- StorageYours
- DeliveryYour edge
- Content keysYour building
Cloud
Fastest to start, and the one where a good month arrives as a bigger invoice.
- Apps and APIsAWS
- EncodingAWS
- StorageAWS
- DeliveryCloudFront
- Content keysYour building
Drawn in full
Two diagrams, both drawn from the running infrastructure, and both open without a call.
Reversible
Moving between the three is a migration, not a rebuild, because the code is identical.
The application, the console and the pipeline are the same build in all three. What changes is which machine each tier runs on, and that is inventory rather than source.