Class ExponentialBackOff

  • All Implemented Interfaces:
    BackOff

    public class ExponentialBackOff
    extends java.lang.Object
    implements BackOff
    Implementation of BackOff that increases the back off period for each retry attempt using a randomization function that grows exponentially.

    nextBackOffMillis() is calculated using the following formula:

     randomized_interval =
     retry_interval * (random value in range [1 - randomization_factor, 1 + randomization_factor])
     

    In other words nextBackOffMillis() will range between the randomization factor percentage below and above the retry interval. For example, using 2 seconds as the base retry interval and 0.5 as the randomization factor, the actual back off period used in the next retry attempt will be between 1 and 3 seconds.

    Note: max_interval caps the retry_interval and not the randomized_interval.

    If the time elapsed since an ExponentialBackOff instance is created goes past the max_elapsed_time then the method nextBackOffMillis() starts returning BackOff.STOP. The elapsed time can be reset by calling reset().

    Example: The default retry_interval is .5 seconds, default randomization_factor is 0.5, default multiplier is 1.5 and the default max_interval is 1 minute. For 10 tries the sequence will be (values in seconds) and assuming we go over the max_elapsed_time on the 10th try:

     request#     retry_interval     randomized_interval
    
     1             0.5                [0.25,   0.75]
     2             0.75               [0.375,  1.125]
     3             1.125              [0.562,  1.687]
     4             1.687              [0.8435, 2.53]
     5             2.53               [1.265,  3.795]
     6             3.795              [1.897,  5.692]
     7             5.692              [2.846,  8.538]
     8             8.538              [4.269, 12.807]
     9            12.807              [6.403, 19.210]
     10           19.210              BackOff.STOP
     

    Implementation is not thread-safe.

    Since:
    1.15
    • Field Detail

      • DEFAULT_INITIAL_INTERVAL_MILLIS

        public static final int DEFAULT_INITIAL_INTERVAL_MILLIS
        The default initial interval value in milliseconds (0.5 seconds).
        See Also:
        Constant Field Values
      • DEFAULT_RANDOMIZATION_FACTOR

        public static final double DEFAULT_RANDOMIZATION_FACTOR
        The default randomization factor (0.5 which results in a random period ranging between 50% below and 50% above the retry interval).
        See Also:
        Constant Field Values
      • DEFAULT_MULTIPLIER

        public static final double DEFAULT_MULTIPLIER
        The default multiplier value (1.5 which is 50% increase per back off).
        See Also:
        Constant Field Values
      • DEFAULT_MAX_INTERVAL_MILLIS

        public static final int DEFAULT_MAX_INTERVAL_MILLIS
        The default maximum back off time in milliseconds (1 minute).
        See Also:
        Constant Field Values
      • DEFAULT_MAX_ELAPSED_TIME_MILLIS

        public static final int DEFAULT_MAX_ELAPSED_TIME_MILLIS
        The default maximum elapsed time in milliseconds (15 minutes).
        See Also:
        Constant Field Values
      • currentIntervalMillis

        private int currentIntervalMillis
        The current retry interval in milliseconds.
      • initialIntervalMillis

        private final int initialIntervalMillis
        The initial retry interval in milliseconds.
      • randomizationFactor

        private final double randomizationFactor
        The randomization factor to use for creating a range around the retry interval.

        A randomization factor of 0.5 results in a random period ranging between 50% below and 50% above the retry interval.

      • multiplier

        private final double multiplier
        The value to multiply the current interval with for each retry attempt.
      • maxIntervalMillis

        private final int maxIntervalMillis
        The maximum value of the back off period in milliseconds. Once the retry interval reaches this value it stops increasing.
      • startTimeNanos

        long startTimeNanos
        The system time in nanoseconds. It is calculated when an ExponentialBackOffPolicy instance is created and is reset when reset() is called.
      • nanoClock

        private final NanoClock nanoClock
        Nano clock.
    • Method Detail

      • reset

        public final void reset()
        Sets the interval back to the initial retry interval and restarts the timer.
        Specified by:
        reset in interface BackOff
      • nextBackOffMillis

        public long nextBackOffMillis()
                               throws java.io.IOException
        Gets the number of milliseconds to wait before retrying the operation or BackOff.STOP to indicate that no retries should be made.

        Example usage:

         long backOffMillis = backoff.nextBackOffMillis();
         if (backOffMillis == Backoff.STOP) {
         // do not retry operation
         } else {
         // sleep for backOffMillis milliseconds and retry operation
         }
         

        This method calculates the next back off interval using the formula: randomized_interval = retry_interval +/- (randomization_factor * retry_interval)

        Subclasses may override if a different algorithm is required.

        Specified by:
        nextBackOffMillis in interface BackOff
        Throws:
        java.io.IOException
      • getRandomValueFromInterval

        static int getRandomValueFromInterval​(double randomizationFactor,
                                              double random,
                                              int currentIntervalMillis)
        Returns a random value from the interval [randomizationFactor * currentInterval, randomizationFactor * currentInterval].
      • getInitialIntervalMillis

        public final int getInitialIntervalMillis()
        Returns the initial retry interval in milliseconds.
      • getRandomizationFactor

        public final double getRandomizationFactor()
        Returns the randomization factor to use for creating a range around the retry interval.

        A randomization factor of 0.5 results in a random period ranging between 50% below and 50% above the retry interval.

      • getCurrentIntervalMillis

        public final int getCurrentIntervalMillis()
        Returns the current retry interval in milliseconds.
      • getMultiplier

        public final double getMultiplier()
        Returns the value to multiply the current interval with for each retry attempt.
      • getMaxIntervalMillis

        public final int getMaxIntervalMillis()
        Returns the maximum value of the back off period in milliseconds. Once the current interval reaches this value it stops increasing.
      • getMaxElapsedTimeMillis

        public final int getMaxElapsedTimeMillis()
        Returns the maximum elapsed time in milliseconds.

        If the time elapsed since an ExponentialBackOff instance is created goes past the max_elapsed_time then the method nextBackOffMillis() starts returning BackOff.STOP. The elapsed time can be reset by calling reset().

      • getElapsedTimeMillis

        public final long getElapsedTimeMillis()
        Returns the elapsed time in milliseconds since an ExponentialBackOff instance is created and is reset when reset() is called.

        The elapsed time is computed using System.nanoTime().

      • incrementCurrentInterval

        private void incrementCurrentInterval()
        Increments the current interval by multiplying it with the multiplier.