What is Rebuffering?
Rebuffering occurs when media playback pauses because the player consumes buffered data faster than replacement data arrives. Playback resumes only after the buffer regains enough content.
How Rebuffering works
A streaming player maintains a time-based reservoir of downloaded, and possibly decoded, media ahead of the playhead. When segment arrival and processing fail to replenish that reservoir before it reaches zero, the player suspends advancement and rebuilds enough headroom to continue. Rebuffering is therefore an outcome across adaptation, CDN delivery, decoding, and device capacity, and its event timing should be captured alongside the rendition and buffer level that preceded it.
Key facts
- 1Startup buffering occurs before playback begins, whereas rebuffering interrupts an active session. Combining them into one delay metric hides two different user-visible failure modes.
- 2High nominal bandwidth does not prevent stalls when throughput varies below the selected rendition, segment requests have long tails, or decoding cannot keep pace with presentation.
- 3Aggressive bitrate reduction can shorten a stall but may create quality oscillation. Adaptation logic needs hysteresis and recent throughput or buffer evidence to avoid repeated switches.
When Rebuffering matters
Track rebuffering frequency and duration when tuning adaptive bitrate selection, segment delivery, and initial buffer size. Choosing a rendition above sustained throughput can cause repeated stalls despite high visual quality.
Common use cases for streaming
These examples cover streaming broadly, not specifically Rebuffering.
- Delivering long-form, episodic, educational, live, or user-generated video over variable networks.
- Providing low-bandwidth through high-resolution renditions from one master.
- Combining captions, alternate audio, encryption, thumbnails, and ad markers with playback media.
Working with streaming
This guidance covers streaming broadly, not just Rebuffering.
An encoder creates several quality levels, and a packager divides them into aligned segments referenced by a manifest. During playback, the client estimates throughput and buffer health, then requests an appropriate segment from one rendition at a time.
Streaming quality depends on the relationship between renditions, segments, manifests, players, and the network. A valid encode can still perform poorly if keyframes are misaligned, the ladder is inefficient, or the player cannot switch cleanly.
What you gain
- Segmented delivery lets playback begin without downloading the entire program.
- Multiple renditions let a player adapt quality as network and device conditions change.
- HTTP-based protocols can reuse ordinary web caching and delivery infrastructure.
What it costs
- Short segments can reduce switching and live latency but increase request and packaging overhead.
- A dense rendition ladder offers finer adaptation while increasing encoding, storage, and cache cost.
- More aggressive quality selection can improve sharpness but raises rebuffering risk on unstable networks.
Before production
- 1Test the rendition ladder on slow, changing, and high-latency connections.
- 2Align segments and keyframes, then validate manifests in the target players.
- 3Measure startup, rebuffering, quality switches, CDN efficiency, and playback failures.