Video Streaming (Netflix-style)
Transcode once, serve adaptive bitrate streams from the edge.
AdvancedMedia
Uploaded masters are transcoded into multiple resolutions/bitrates, stored, and served through a CDN using adaptive bitrate streaming so playback quality adjusts to each viewer's network in real time.
When to use it
- Large catalog of on-demand video watched by a geographically distributed audience
- Playback quality needs to adapt automatically to fluctuating client bandwidth
Trade-offs
- Transcoding a large catalog into many renditions is computationally and storage expensive
- CDN cache misses on cold/long-tail content still hit origin at full cost
Components used
Object StorageBatch / Scheduled JobCDNDesktop AppManaged App ServiceDocument DatabaseLLM / Model Endpoint
How it works
- Video is transcoded once into a ladder of resolutions and bitrates, then segmented into small chunks with a manifest describing them.
- The player measures available bandwidth and requests the next chunk at whichever quality it can sustain, switching mid-stream without interruption.
- Chunks are served from CDN edges, so the origin serves almost no viewer traffic.
Used in the wild
- Subscription video-on-demand catalogues.
- Course and training platforms.
- Any long-form video where buffering is the primary quality complaint.
Good to know
- Adaptive bitrate is why video degrades in quality instead of stalling. The player is continuously making a bandwidth bet, one segment at a time.
- Netflix places its own appliances inside ISP networks so popular content is served from within the ISP entirely. At peak, video streaming is a large share of all consumer internet traffic, and this is how it stays viable.
Related system design
Video Upload & Transcode (YouTube-style)
Accept large uploads, process asynchronously, notify when ready.
Live Streaming Platform
Ingest a live feed, transcode in near real time, and fan out to viewers at the edge.
File Sync Service (Dropbox-style)
Chunk, deduplicate and sync files across devices with conflict resolution.
Collaborative Document Editor (Google Docs-style)
Multiple users edit the same document in real time via operational transforms.