Package org.opennms.core.ipc.sink.api

Interface AggregationPolicy<S,​T,​U>

  • Type Parameters:
    S - type of message that will be sent by the producers
    T - type of message that will be received by the consumers
    U - intermerdiary accumulator type used to aggregate the messages
    All Known Implementing Classes:
    ArrayListAggregationPolicy, IdentityAggregationPolicy, MappingAggregationPolicy
    public interface AggregationPolicy<S,​T,​U>
    Defines how messages will be aggregated. When a SinkModule defines a AggregationPolicy, messages will be aggregated by accumulators, which are keyed based on the object returned by key(Object). The aggregation function aggregate(Object, Object) is called to create accumulators and add messages to existing accumulators. The completion size and completion interval options determine the conditions under which the buckets will be dispatched.
    Author:
    jwhite

    • Method Summary

      All Methods Instance Methods Abstract Methods 
      Modifier and Type Method Description
      U aggregate​(U accumulator, S newMessage)
      Aggregate the given message into an existing accumulator, or create a new accumulator if no accumulator exists.
      T build​(U accumulator)
      Build the resulting message from the accumulator.
      int getCompletionIntervalMs()
      Maximum number of milliseconds for which buckets should continue accumulating messages after creation.
      int getCompletionSize()
      Maximum number of messages to be added to a bucket before dispatching.
      java.lang.Object key​(S message)
      Calculate a key for the given message.

    • Method Detail

      • getCompletionSize

        int getCompletionSize()
        Maximum number of messages to be added to a bucket before dispatching. If this value is <= 1, the buckets should be dispatched immediately after adding a single element.
        Returns:
        maximum number of messages per bucket

      • getCompletionIntervalMs

        int getCompletionIntervalMs()
        Maximum number of milliseconds for which buckets should continue accumulating messages after creation. If a bucket has been created for longer than this interval, it will be dispatched regardless of it's current size. Values <= 0 will disable periodic flushing and buckets will only be dispatched once they have reached the maximum size.
        Returns:
        number of milliseconds to keep a bucket before dispatching

      • key

        java.lang.Object key​(S message)
        Calculate a key for the given message. Objects with the same key will be aggregated together. Returned values should non-null.
        Parameters:
        message - the message
        Returns:
        the message's key

      • aggregate

        U aggregate​(U accumulator,
            S newMessage)
        Aggregate the given message into an existing accumulator, or create a new accumulator if no accumulator exists.
        Parameters:
        accumulator - the existing accumulator, or null if a new accumulator should be created
        newMessage - the message to aggregate
        Returns:
        the new or updated accumulator

      • build

        T build​(U accumulator)
        Build the resulting message from the accumulator.
        Parameters:
        accumulator - an existing accumulator
        Returns:
        the aggregated message to dispatch