Full Jitter
A jitter scheme that first computes the exponential backoff cap and then takes a uniformly random delay within the interval [0, cap], using the full window range to spread client retry load to the greatest extent.
Timeline (last 90 days)
AWS's "Full Jitter" strategy recommends a retry wait time of random(0, min(cap, base * 2^attempt)).
The AWS engineering blog compared several implementations including pure exponential backoff, full jitter, equal jitter, and decorrelated jitter; experimental results showed that Full Jitter achieves the optimal balance between throughput and server-side pressure.
Full jitter sacrifices retry latency for individual clients, but significantly reduces the temporal synchronization among the client population, most effectively smoothing out traffic spikes.
All Facts (3)
AWS's "Full Jitter" strategy recommends a retry wait time of random(0, min(cap, base * 2^attempt)).
50%UnverifiedThe AWS engineering blog compared several implementations including pure exponential backoff, full jitter, equal jitter, and decorrelated jitter; experimental results showed that Full Jitter achieves the optimal balance between throughput and server-side pressure.
50%UnverifiedFull jitter sacrifices retry latency for individual clients, but significantly reduces the temporal synchronization among the client population, most effectively smoothing out traffic spikes.
50%