Mailing-List: contact dev-help@pig.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@pig.apache.org
Date: Sun, 4 Dec 2011 06:06:40 +0000 (UTC)
From: "Jie Li (Created) (JIRA)" <jira@apache.org>
To: pig-dev@hadoop.apache.org
Message-ID: 
 <1356762751.38200.1322978800113.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Created] (PIG-2395) Pig doesn't do early projection for
 some scripts
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Pig doesn't do early projection for some scripts
------------------------------------------------

                 Key: PIG-2395
                 URL: https://issues.apache.org/jira/browse/PIG-2395
             Project: Pig
          Issue Type: Improvement
    Affects Versions: 0.9.1
         Environment: Linux
            Reporter: Jie Li


For some scripts Pig doesn't do early projection, e.g. dropping non-necessary fields as soon as possible. This is observed in two ways: 1) the output doesn't contain INFO like "ColumnPruneVisitor - Columns pruned for xxx: $0, $1"; 2) the job has as much or more local IO (see FILE_BYTES_READ and FILE_BYTES_WRITTEN) as HDFS IO (see HDFS_BYTES_READ).

One example where Pig should figure out A's fields c~n can be dropped before the COGROUP.

A = load '/tmp/A' USING PigStorage('|') as (a,b,c,d,e,f,g,h,i,j,k,l,m,n);
B = load '/tmp/B' USING PigStorage('|') as (a);
COG = cogroup A by a, B by a;
out = foreach COG generate SUM(A.b) as sum;
store out into '/tmp/out' USING PigStorage('|');

Another similar example involves a GROUP operator.

While Pig is able and assumed to do early projection in most cases, this inconsistency hurts the performance badly. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira