[ 
https://issues.apache.org/jira/browse/FLINK-9423?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16493602#comment-16493602
 ] 

ASF GitHub Bot commented on FLINK-9423:
---------------------------------------

Github user tillrohrmann commented on a diff in the pull request:

    https://github.com/apache/flink/pull/6062#discussion_r191423366
  
    --- Diff: 
flink-streaming-java/src/main/java/org/apache/flink/streaming/api/operators/InternalTimerHeap.java
 ---
    @@ -0,0 +1,511 @@
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + * http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +package org.apache.flink.streaming.api.operators;
    +
    +import org.apache.flink.annotation.VisibleForTesting;
    +import org.apache.flink.runtime.state.KeyGroupRange;
    +import org.apache.flink.runtime.state.KeyGroupRangeAssignment;
    +
    +import javax.annotation.Nonnegative;
    +import javax.annotation.Nonnull;
    +import javax.annotation.Nullable;
    +
    +import java.util.ArrayList;
    +import java.util.Arrays;
    +import java.util.Collection;
    +import java.util.Collections;
    +import java.util.Comparator;
    +import java.util.HashMap;
    +import java.util.Iterator;
    +import java.util.List;
    +import java.util.NoSuchElementException;
    +import java.util.Queue;
    +import java.util.Set;
    +
    +import static org.apache.flink.util.Preconditions.checkArgument;
    +
    +/**
    + * A heap-based priority queue for internal timers. This heap is supported 
by hash sets for fast contains
    + * (de-duplication) and deletes. The heap implementation is a simple 
binary tree stored inside an array. Element indexes
    + * in the heap array start at 1 instead of 0 to make array index 
computations a bit simpler in the hot methods.
    + *
    + * <p>Possible future improvements:
    + * <ul>
    + *  <li>We could also implement shrinking for the heap and the 
deduplication maps.</li>
    + *  <li>We could replace the deduplication maps with more efficient custom 
implementations. In particular, a hash set
    + * would be enough if it could return existing elements on unsuccessful 
adding, etc..</li>
    + * </ul>
    + *
    + * @param <K> type of the key of the internal timers managed by this 
priority queue.
    + * @param <N> type of the namespace of the internal timers managed by this 
priority queue.
    + */
    +public class InternalTimerHeap<K, N> implements 
Queue<TimerHeapInternalTimer<K, N>>, Set<TimerHeapInternalTimer<K, N>> {
    --- End diff --
    
    That way we could also save us from implementing some methods like 
`toArray` or `iterator`.


> Implement efficient deletes for heap based timer service
> --------------------------------------------------------
>
>                 Key: FLINK-9423
>                 URL: https://issues.apache.org/jira/browse/FLINK-9423
>             Project: Flink
>          Issue Type: Improvement
>          Components: Streaming
>    Affects Versions: 1.5.0
>            Reporter: Stefan Richter
>            Assignee: Stefan Richter
>            Priority: Major
>             Fix For: 1.6.0
>
>
> The current data structures in the `HeapInternalTimerService` are not able to 
> support efficient timer deletes, the complexity is currently O\(n\), where n 
> is the number of registered timers.
>  
> We can keep track of timer's positions in the priority queue and (in 
> combination with the already existing set/map) have a more efficient 
> algorithm for deletes.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to