sqoop-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jarek Jarcec Cecho (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SQOOP-1968) Optimize schema operation in getMatchingData of NameMatcher
Date Tue, 06 Jan 2015 10:12:34 GMT

    [ https://issues.apache.org/jira/browse/SQOOP-1968?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14265926#comment-14265926
] 

Jarek Jarcec Cecho commented on SQOOP-1968:
-------------------------------------------

I'm generally +1 on the patch, I have two nits:

* Nit: There are unused imports in {{NameMatcher.java}}, can we remove them?
* In file {{SchemaFixture.java}} I would suggest to rename methods {{createRandomSchema}}
and {{createRandomSchema}} to not contain word "random" as the schema doesn't seem to be randomized.

> Optimize schema operation in getMatchingData of NameMatcher
> -----------------------------------------------------------
>
>                 Key: SQOOP-1968
>                 URL: https://issues.apache.org/jira/browse/SQOOP-1968
>             Project: Sqoop
>          Issue Type: Improvement
>          Components: connectors/generic
>    Affects Versions: 2.0.0
>            Reporter: Jerry Chen
>            Assignee: Qian Xu
>         Attachments: SQOOP-1968.patch
>
>
> Two performance issues found in the Matcher implementations.
> 1. In getMatchingData  of NameMatcher, the following code block of building a HashMap
will not change across different getMatchingData  calls. The HashMap can build only once in
Constructor.
> {code}
> HashMap<String,Column> colNames = new HashMap<String, Column>();
> for (Column fromCol: getFromSchema().getColumnsArray()) {
>   colNames.put(fromCol.getName(), fromCol);
> }
> {code}
> 2. In getMatchingData  of NameMatcher, indexOf of a List implementation is not efficient.
It usually involves a loop for finding the object and return the index. To improve, we can
simple store the index in the above HashMap and retrieve the index by HashMap lookup directly
> {code}
> int fromIndex = getFromSchema().getColumnsList().indexOf(fromCol);
> {code}
> These performance problems are critical because getMatchingData is repeatedly calling
for each record.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message