Mailing-List: contact avro-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: avro-dev@hadoop.apache.org
Message-ID: <2134591785.1262289809394.JavaMail.jira@brutus.apache.org>
Date: Thu, 31 Dec 2009 20:03:29 +0000 (UTC)
From: "Philip Zeyliger (JIRA)" <jira@apache.org>
To: avro-dev@hadoop.apache.org
Subject: [jira] Updated: (AVRO-275) Histogram class to keep RPC timing stats
In-Reply-To: <214673257.1262284289389.JavaMail.jira@brutus.apache.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/AVRO-275?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Philip Zeyliger updated AVRO-275:
---------------------------------

    Attachment: AVRO-275.patch.txt

Attaching a patch.

I went for a non-adaptive approach: you have to specify the formula (a "segmenter" implementation: I don't much like the name) by which to bucket your values ahead of time, as well as the number of buckets.  I wrote one implementation of this formula (using a TreeMap), and linear and exponential versions make sense too.

This may be more convoluted than necessary: I support generic buckets, not just numbers.

I looked around a bit on whether I should have borrowed other Apache-licensed code.  http://code.google.com/p/hist4j/ has an adaptive approach.  Apache commons math has summary statistics code.

> Histogram class to keep RPC timing stats
> ----------------------------------------
>
>                 Key: AVRO-275
>                 URL: https://issues.apache.org/jira/browse/AVRO-275
>             Project: Avro
>          Issue Type: New Feature
>          Components: java
>            Reporter: Philip Zeyliger
>            Assignee: Philip Zeyliger
>            Priority: Minor
>         Attachments: AVRO-275.patch.txt
>
>
> Class to keep a histogram of RPC timinigs for AVRO-270.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.