Developers

Random Jitter

Random jitter is a technique used in conjunction with exponential backoff to introduce a random delay within a defined range. Instead of a fixed, exponentially increasing delay, the actual waiting time is randomized. For example, if the calculated exponential backoff delay is 4 seconds, a random jitter algorithm might choose a delay anywhere between 0 and 4 seconds. This approach helps to prevent a “thundering herd” problem, where multiple clients, after experiencing a failure, all retry at the exact same moment.

Example implementation

Pseudo-code

define max_retries
define initial_delay

function retry(operation, max_retries, initial_delay)
    // Initialize the attempt counter and the current delay
    attempt <- 0
    delay <- initial_delay

    while attempt < max_retries do
        try
            // Attempt to perform the operation
            operation()
            // If successful, exit the function
            return
        catch
            // If the operation fails, increment the attempt counter
            attempt <- attempt + 1
            // Calculate the delay with a random jitter component
            jittered_delay <- delay + random(0, delay)
            // Pause for the jittered delay period
            sleep jittered_delay
            // Double the base delay for the next retry (exponential backoff)
            delay <- delay * 2
    end while
    // If all retries fail, the function will finish here
end function

Simulation