Request latency
Request latency is the time between a player asking a server for a piece of video and the first byte of the reply arriving. High latency usually points at a far-away or overloaded CDN edge rather than a slow connection, and it adds directly to startup time and seek time.
Why it matters
Every segment request waits on request latency before any video arrives, so slow responses add directly to startup time, seek time and stalls. Measuring it by CDN and region shows where an edge is underperforming and gives evidence for a multi-CDN setup or a contract conversation. Both the evidence and the multi-CDN setup come together on FastPix: sessions are routed across multiple CDNs, and Video Data records which CDN served every view.
How it works
- The player sends a request to the CDN for a playlist or video segment.
- It notes the time the request was sent.
- It notes when the first byte of the response arrives.
- The gap between the two is the request latency, reported as an average and a maximum per view.
What is a good request latency for video?
A good segment request latency is generally under about 100 milliseconds when viewers are near a healthy CDN edge. Values in the hundreds of milliseconds suggest a distant or overloaded edge, and they add up quickly because a player makes many requests per minute. Compare by region, since viewers far from edge locations will always see higher numbers.
Request latency vs throughput: what is the difference?
Request latency is how long the first byte takes to arrive; throughput is how fast the rest of the data downloads. High latency with good throughput points to distance or a slow server response. Low latency with poor throughput points to a congested or weak connection. Video needs both to be healthy for fast starts and smooth playback.
Where it's used
Video platforms and live streamers break request latency down by CDN and region to see which edge locations are slow and whether a second CDN would help.
Where it usually goes wrong, and how FastPix handles it
Teams usually see request latency only as an overall average, which hides the slow requests that cause stalls and hides which CDN or region is responsible.
Average and maximum request latency come from each segment request in FastPix Video Data, tagged with the CDN that served the view, so latency can be compared by CDN, country and ISP. The result is a clear case for which edge or CDN to fix, backed by real viewer data rather than one average.
By the numbers
- In the browser Resource Timing API, responseStart marks the moment the first byte of a response arrives from the server, cache or local resource. (PerformanceResourceTiming: responseStart property (MDN))
