[
https://issues.apache.org/jira/browse/FLINK-1297?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14503979#comment-14503979
]
ASF GitHub Bot commented on FLINK-1297:
---------------------------------------
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/605#discussion_r28739811
--- Diff:
flink-core/src/main/java/org/apache/flink/statistics/OperatorStatistics.java ---
@@ -0,0 +1,154 @@
+/*
+ * 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.statistics;
+
+import
com.clearspring.analytics.stream.cardinality.CardinalityMergeException;
+import com.clearspring.analytics.stream.cardinality.HyperLogLog;
+import com.clearspring.analytics.stream.cardinality.ICardinality;
+import com.clearspring.analytics.stream.cardinality.LinearCounting;
+import org.apache.flink.statistics.heavyhitters.IHeavyHitter;
+import org.apache.flink.statistics.heavyhitters.LossyCounting;
+import org.apache.flink.statistics.heavyhitters.CountMinHeavyHitter;
+import org.apache.flink.statistics.heavyhitters.HeavyHitterMergeException;
+
+import java.io.Serializable;
+import java.util.Map;
+
+/**
+ * Data structure that encapsulates statistical information of data that
has only been processed by one pass
+ * This statistical information is meant to help determine the
distribution of the data that has been processed
+ * in an operator so that we can determine if it is necessary to
repartition the data
+ *
+ * The statistics to be gathered are configurable and represented by a
{@link OperatorStatisticsConfig} object.
+ *
+ * The information encapsulated in this class is min, max, a structure
enabling estimation of count distinct and a
+ * structure holding the heavy hitters along with their frequency.
+ *
+ */
+public class OperatorStatistics implements Serializable {
+
+ OperatorStatisticsConfig config;
+
+ Object min;
+ Object max;
+ ICardinality countDistinct;
+ IHeavyHitter heavyHitter;
+ long cardinality = 0;
+
+ public OperatorStatistics(OperatorStatisticsConfig config) {
+ this.config = config;
+ if
(config.countDistinctAlgorithm.equals(OperatorStatisticsConfig.CountDistinctAlgorithm.LINEAR_COUNTING))
{
+ countDistinct = new
LinearCounting(OperatorStatisticsConfig.COUNTD_BITMAP_SIZE);
+ }
+
if(config.countDistinctAlgorithm.equals(OperatorStatisticsConfig.CountDistinctAlgorithm.HYPERLOGLOG)){
+ countDistinct = new
HyperLogLog(OperatorStatisticsConfig.COUNTD_LOG2M);
+ }
+ if
(config.heavyHitterAlgorithm.equals(OperatorStatisticsConfig.HeavyHitterAlgorithm.LOSSY_COUNTING)){
+ heavyHitter =
+ new
LossyCounting(OperatorStatisticsConfig.HEAVY_HITTER_FRACTION,
OperatorStatisticsConfig.HEAVY_HITTER_ERROR);
+ }
+ if
(config.heavyHitterAlgorithm.equals(OperatorStatisticsConfig.HeavyHitterAlgorithm.COUNT_MIN_SKETCH)){
+ heavyHitter =
+ new
CountMinHeavyHitter(OperatorStatisticsConfig.HEAVY_HITTER_FRACTION,
+
OperatorStatisticsConfig.HEAVY_HITTER_ERROR,
+
OperatorStatisticsConfig.HEAVY_HITTER_CONFIDENCE,
+
OperatorStatisticsConfig.HEAVY_HITTER_SEED);
+ }
+ }
+
+ public void process(Object tupleObject){
--- End diff --
It looks like this method expected to be called for each passing element if
statistics collection is enabled. This could add significant processing
overhead. Would it make sense to add an optional skip interval n that processes
only every n-th element (except from increasing the `cardinality` counter.
Also it makes sense to perform as many checks outside of the function as
possible. The type of the data elements is known before execution. So it can be
checked if a data type allows statistics collection at program composition or
optimization time.
> Add support for tracking statistics of intermediate results
> -----------------------------------------------------------
>
> Key: FLINK-1297
> URL: https://issues.apache.org/jira/browse/FLINK-1297
> Project: Flink
> Issue Type: Improvement
> Components: Distributed Runtime
> Reporter: Alexander Alexandrov
> Assignee: Alexander Alexandrov
> Fix For: 0.9
>
> Original Estimate: 1,008h
> Remaining Estimate: 1,008h
>
> One of the major problems related to the optimizer at the moment is the lack
> of proper statistics.
> With the introduction of staged execution, it is possible to instrument the
> runtime code with a statistics facility that collects the required
> information for optimizing the next execution stage.
> I would therefore like to contribute code that can be used to gather basic
> statistics for the (intermediate) result of dataflows (e.g. min, max, count,
> count distinct) and make them available to the job manager.
> Before I start, I would like to hear some feedback form the other users.
> In particular, to handle skew (e.g. on grouping) it might be good to have
> some sort of detailed sketch about the key distribution of an intermediate
> result. I am not sure whether a simple histogram is the most effective way to
> go. Maybe somebody would propose another lightweight sketch that provides
> better accuracy.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
