Reliability
Reliable video hosting you can verify for yourself
Anyone can print a percentage on a marketing page. We would rather hand you the measurements: how long your viewers wait for the first frame, how often playback rebuffers, and how often it errors — reported per account from real playback. Combined with adaptive delivery that steps down a rendition instead of stopping, that gives you reliability you can check rather than reliability you have to take on trust.
How it works
How playback actually reaches a viewer
Reliability in video is mostly about degrading well. These are the mechanisms that decide whether a bad minute is invisible or fatal.
- Adaptive HLS with multiple renditionsEvery video
- One upload becomes a set of renditions. When a viewer’s connection weakens the player steps down to a lower one and keeps playing, which is why adaptive delivery matters far more to a real audience than a headline bitrate does.
- Segmented deliveryEvery video
- Video is delivered as short segments rather than one long file, so a single failed request retries in isolation instead of taking the whole playback with it.
- Playback quality measured from real sessionsEvery plan
- The platform records the events that actually describe reliability — bootstrap, manifest request, first frame, rebuffer, heartbeat, error — and reports them per account. You get a playback quality report rather than a promise.
- Per-video status you can poll or subscribe toEvery plan
- Each video exposes its processing status through the API, and webhook endpoints tell your system when state changes. Your publishing workflow does not have to guess whether a video is ready.
- Webhook delivery with retriesEvery plan
- Webhook deliveries are recorded and can be retried, and endpoint secrets can be rotated. If your integration is briefly unavailable, the event is not simply lost.
- Resumable uploadsEvery plan
- A dropped connection during a large upload resumes rather than restarting, which is the reliability property that matters most on migration day.
Practical
How to build a page that survives a bad minute
Most visible video failures are page-design failures rather than delivery failures. Five things worth doing whichever host you choose.
- 01
1. Never let the video block the page
- 02
2. Give the player a reserved space
- 03
3. Provide the text alternative
- 04
4. Watch your own quality report
- 05
5. Get the commitments you need in writing
Where it matters most
Libraries where playback quality is the product
Course platforms
A rebuffering lesson is a refund request. Adaptive renditions plus a quality report you can watch is the combination that keeps cohorts moving.
Product and conversion pages
A demo that stalls costs a sale. Keep the page usable without the video, and measure first-frame timing where it counts.
Internal communication
Corporate networks and proxies are unusually hostile to video. Segmented adaptive delivery and the error report tell you where the friction actually is.
Recorded events
Long recordings are where rebuffering compounds. Chapters plus a lower rendition keep a two-hour session watchable on a weak connection.
Systems-driven publishing
Per-video status and retried webhooks mean an automated pipeline does not publish a page pointing at a video that is not ready.
Teams that report on quality
Because the measurements come back through the API, playback quality can sit on your own dashboard next to everything else you monitor.
Questions
Reliability questions, answered straight
How can I evaluate playback reliability?
Use the playback quality report for first-frame timing, rebuffering and errors from your own audience. If your organisation requires contractual availability or support commitments, share those requirements with our team and evaluate the written commercial terms alongside the measured playback data.
What playback measurements can I see?
Playback quality is recorded from real sessions — bootstrap, manifest request, first frame, rebuffer, heartbeat and error events — and reported per account. That tells you how long viewers wait, how often playback stalls and how often it fails, which is what reliability means in practice.
What happens when a viewer has a poor connection?
Adaptive HLS steps down to a lower rendition and keeps playing rather than stopping. Because delivery is segmented, a single failed request retries on its own instead of ending the session. Most viewers on a weak connection see a softer picture rather than an error.
How do I stop a video problem breaking my page?
Render the page content first and never make it depend on the player, reserve a fixed aspect-ratio space so the layout does not jump, and provide captions or a transcript so someone who cannot play the video still gets the information. Those three habits prevent most of the damage a bad minute can do.
Can I get a written commitment?
Ask us. Tell us which commitments your business or procurement process actually needs, and we will answer against them plainly and in writing. Rely on that document rather than on any page on this website.
How will my systems know if a video is not ready?
Each video exposes its processing status through the API, and webhook endpoints notify your system when state changes. Deliveries are recorded and can be retried, and endpoint secrets can be rotated, so an automated pipeline does not publish a page pointing at a video that has not finished processing.
Measure it, then decide
Publish a handful of real videos, watch the playback quality report for your own audience, and ask us for whatever commitments your business needs in writing. That is a firmer basis than any percentage.