What you gain
No hardware, no data-centre relationship, and a launch that does not wait on procurement. Capacity is available the moment you need it.
All of it on AWS — engineered, not lifted
Sometimes there is no rack and there is not going to be one. This pattern puts every tier on AWS, but designs the cost controls in from the first commit rather than discovering them in month four. Right-sized instances, an egress path chosen deliberately, storage classes that match access reality, and commitments bought against measured usage instead of hope.
How it fits together
Every tier named, every boundary drawn. Blue is capacity you own; grey is capacity you rent.
Web · mobile · television
- Web player
- Mobile apps
- TV apps
- Playback telemetry
The tier to watch on the invoice
- CDN distribution
- Signed URLs / tokens
- Load balancing
- WAF & rate limits
Containers, right-sized
- Catalogue & search
- Auth & entitlements
- Billing & payments
- Analytics ingest
Storage class matched to access
- Encoding jobs
- Object storage
- Managed database
- Secrets manager
No console clicking, ever
- Terraform
- GitOps pipeline
- Budgets & anomaly alerts
- Commitment strategy
Tier by tier
| Concern | Default outcome | Engineered outcome |
|---|---|---|
| Egress | Metered on every segment, unbounded | Cache policy and commit pricing chosen against measured traffic |
| Compute | Oversized instances, idle overnight | Right-sized, scaled to real concurrency |
| Storage | Everything in the hot class forever | Lifecycle rules matched to actual access curves |
| Transcoding | Per-minute managed service | Owned GPU instances, batched |
| Commitments | Bought on a guess, or not at all | Bought against measured baseline |
| Change control | Console clicks nobody can replay | Terraform plan in a merge request |
What it buys and what it costs
Every architecture pays for something. Here is the bill for this one.
What you take on
A bill that grows with success. That is survivable when it is engineered and monitored; it becomes existential when nobody owns it.
The usual next step
Most services that grow on this pattern eventually move delivery and origin onto their own metal. Because everything is Terraform, that migration is a change of target, not a rewrite.
When this is the right pattern
No hardware, a hard launch date, or a compliance position that requires a hyperscaler. You still want the bill engineered.
The other two patterns
Most evaluations end up comparing all three. They are all delivered, and none of them is the default answer.
Which one fits your estate?
Send what you have — hardware, cloud account, current bill, launch date. The answer is usually one of these three, and occasionally it is that you should change nothing.