[
https://issues.apache.org/jira/browse/HADOOP-19729?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18033640#comment-18033640
]
ASF GitHub Bot commented on HADOOP-19729:
-----------------------------------------
manika137 commented on code in PR #8043:
URL: https://github.com/apache/hadoop/pull/8043#discussion_r2470581573
##########
hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azurebfs/services/AbfsTailLatencyTracker.java:
##########
@@ -0,0 +1,159 @@
+/**
+ * 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.hadoop.fs.azurebfs.services;
+
+import java.util.HashMap;
+import java.util.Map;
+import java.util.concurrent.Executors;
+import java.util.concurrent.ScheduledExecutorService;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.locks.ReentrantLock;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import org.apache.hadoop.fs.azurebfs.AbfsConfiguration;
+
+/**
+ * Account Specific Latency Tracker.
+ * This class tracks the latency of various operations like read, write etc
for a single account.
+ * It maintains a sliding window histogram for each operation type to analyze
latency patterns over time.
+ */
+public class AbfsTailLatencyTracker {
+
+ private static final Logger LOG = LoggerFactory.getLogger(
+ AbfsTailLatencyTracker.class);
+ private static AbfsTailLatencyTracker singleton;
+ private static final ReentrantLock LOCK = new ReentrantLock();
+ private static final int HISTOGRAM_MAX_VALUE = 60_000;
+ private static final int HISTOGRAM_SIGNIFICANT_FIGURES = 3;
+ private final Map<AbfsRestOperationType, SlidingWindowHdrHistogram>
+ operationLatencyMap = new HashMap<>();
+ private final AbfsConfiguration configuration;
+
+ /**
+ * Constructor to initialize the latency tracker with configuration.
+ * @param abfsConfiguration Configuration settings for latency tracking.
+ */
+ public AbfsTailLatencyTracker(AbfsConfiguration abfsConfiguration) {
+ this.configuration = abfsConfiguration;
+ ScheduledExecutorService histogramRotatorThread =
Executors.newSingleThreadScheduledExecutor(
+ r -> {
+ Thread t = new Thread(r, "Histogram-Rotator-Thread");
+ t.setDaemon(true);
+ return t;
+ });
+ long rotationInterval =
configuration.getTailLatencyAnalysisWindowInMillis()
+ / configuration.getTailLatencyAnalysisWindowGranularity();
+ histogramRotatorThread.scheduleAtFixedRate(this::rotateHistograms,
+ rotationInterval, rotationInterval, TimeUnit.MILLISECONDS);
+
+
+ ScheduledExecutorService tailLatencyComputationThread =
Executors.newSingleThreadScheduledExecutor(
+ r -> {
+ Thread t = new Thread(r, "Tail-Latency-Computation-Thread");
+ t.setDaemon(true);
+ return t;
+ });
+
+ long computationalInterval =
configuration.getTailLatencyPercentileComputationIntervalInMillis();
+ tailLatencyComputationThread.scheduleAtFixedRate(this::computePercentiles,
+ computationalInterval, computationalInterval, TimeUnit.MILLISECONDS);
+ }
+
+ /**
+ * Rotates all histograms to ensure they reflect the most recent latency
data.
+ * This method is called periodically based on the configured rotation
interval.
+ */
+ private void rotateHistograms() {
+ for (SlidingWindowHdrHistogram histogram : operationLatencyMap.values()) {
+ histogram.rotateIfNeeded();
+ }
+ }
+
+ /**
+ * Computes the tail latency percentiles for all operation types.
+ * This method is called periodically based on the configured computation
interval.
+ */
+ private void computePercentiles() {
+ for (SlidingWindowHdrHistogram histogram : operationLatencyMap.values()) {
+ histogram.computeLatency();
+ }
+ }
+
+ /**
+ * Creates a singleton object of the {@link SlidingWindowHdrHistogram}.
+ * which is shared across all filesystem instances.
+ * @param abfsConfiguration configuration set.
+ * @return singleton object of intercept.
+ */
+ static AbfsTailLatencyTracker initializeSingleton(AbfsConfiguration
abfsConfiguration) {
+ if (singleton == null) {
+ LOCK.lock();
+ try {
+ if (singleton == null) {
+ singleton = new AbfsTailLatencyTracker(abfsConfiguration);
Review Comment:
we could log the initialization with the granularity etc configs here
> ABFS: [Perf] Network Profiling of Tailing Requests and Killing Bad
> Connections Proactively
> ------------------------------------------------------------------------------------------
>
> Key: HADOOP-19729
> URL: https://issues.apache.org/jira/browse/HADOOP-19729
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/azure
> Affects Versions: 3.4.2
> Reporter: Anuj Modi
> Assignee: Anuj Modi
> Priority: Major
> Labels: pull-request-available
>
> It has been observed that certain requests taking more time than expected to
> complete hinders the performance of whole workload. Such requests are known
> as tailing requests. They can be taking more time due to a number of reasons
> and the prominent among them is a bad network connection. In Abfs driver we
> cache network connections and keeping such bad connections in cache and
> reusing them can be bad for perf.
> In this effort we try to identify such connections and close them so that new
> good connetions can be established and perf can be improved. There are two
> parts of this effort.
> # Identifying Tailing Requests: This involves profiling all the network
> calls and getting percentiles value optimally. By default we consider p99 as
> the tail latency and all the future requests taking more than tail latency
> will be considere as Tailing requests.
> # Proactively Killing Socket Connections: With Apache client, we can now
> kill the socket connection and fail the tailing request. Such failures will
> not be thrown back to user and retried immediately without any sleep but from
> another socket connection.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
