[
https://issues.apache.org/jira/browse/HDFS-16949?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17701902#comment-17701902
]
ASF GitHub Bot commented on HDFS-16949:
---------------------------------------
mkuchenbecker commented on code in PR #5486:
URL: https://github.com/apache/hadoop/pull/5486#discussion_r1140468345
##########
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/metrics2/lib/MutableQuantiles.java:
##########
@@ -104,8 +105,7 @@ public MutableQuantiles(String name, String description,
String sampleName,
String.format(descTemplate, percentile));
}
- estimator = new SampleQuantiles(quantiles);
-
+ estimator = inverseQuantiles ? new InverseQuantiles(quantiles) : new
SampleQuantiles(quantiles);
Review Comment:
Is estimator variable used?
##########
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/metrics2/util/InverseQuantiles.java:
##########
@@ -0,0 +1,44 @@
+package org.apache.hadoop.metrics2.util;
+
+import org.apache.hadoop.util.Preconditions;
+import java.util.ListIterator;
+
+public class InverseQuantiles extends SampleQuantiles{
+
+ public InverseQuantiles(Quantile[] quantiles) {
+ super(quantiles);
+ }
+
+
+ /**
+ * Get the estimated value at the inverse of the specified quantile.
+ * Eg: return the value at (1 - 0.99)*count position for quantile 0.99.
+ * When count is 100, quantile 0.99 is desired to return the value at the
1st position
+ *
+ * @param quantile Queried quantile, e.g. 0.50 or 0.99.
+ * @return Estimated value at the inverse position of that quantile.
+ */
+ long query(double quantile) {
Review Comment:
- Minimal change here is to inverse list-order traversal.
- The for loop is performing a Map operation.
The DRY method would be to encapsulate the for loop to a protected function
"getQuantilesFromSamples".
> Update ReadTransferRate to ReadLatencyPerGB for effective percentile metrics
> ----------------------------------------------------------------------------
>
> Key: HDFS-16949
> URL: https://issues.apache.org/jira/browse/HDFS-16949
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: datanode
> Reporter: Ravindra Dingankar
> Assignee: Ravindra Dingankar
> Priority: Minor
> Labels: pull-request-available
> Fix For: 3.3.0, 3.4.0
>
>
> HDFS-16917 added ReadTransferRate quantiles to calculate the rate which data
> is read per unit of time.
> With percentiles the values are sorted in ascending order and hence for the
> transfer rate p90 gives us the value where 90 percent rates are lower
> (worse), p99 gives us the value where 99 percent values are lower (worse).
> Note that value(p90) < p(99) thus p99 is a better transfer rate as compared
> to p90.
> However as the percentile increases the value should become worse in order to
> know how good our system is.
> Hence instead of calculating the data read transfer rate, we should calculate
> it's inverse. We will instead calculate the time taken for a GB of data to be
> read. ( seconds / GB )
> After this the p90 value will give us 90 percentage of total values where the
> time taken is less than value(p90), similarly for p99 and others.
> Also p(90) < p(99) and here p(99) will become a worse value (taking more time
> each byte) as compared to p(90)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
