asterixdb-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF subversion and git services (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ASTERIXDB-1077) Inverted index tests are slow compared to others
Date Thu, 08 Oct 2015 20:25:27 GMT

    [ https://issues.apache.org/jira/browse/ASTERIXDB-1077?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14949325#comment-14949325
] 

ASF subversion and git services commented on ASTERIXDB-1077:
------------------------------------------------------------

Commit aaa888ba68568eea5b32a9afc2dfd885263d7aa9 in incubator-asterixdb-hyracks's branch refs/heads/master
from [~wangsaeu]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-asterixdb-hyracks.git;h=aaa888b ]

ASTERIXDB-1077: Fixed - Inverted index tests are slow compared to others

- Reduced the number of searches during the inverted index searches
- Reduced the number of Jaccard search modifiers

Change-Id: I1d8fc1d30055798fee7f3f23d08a57af867e3a3c
Reviewed-on: https://asterix-gerrit.ics.uci.edu/442
Tested-by: Jenkins <jenkins@fulliautomatix.ics.uci.edu>
Reviewed-by: Young-Seok Kim <kisskys@gmail.com>
Reviewed-by: Ian Maxon <imaxon@apache.org>


> Inverted index tests are slow compared to others
> ------------------------------------------------
>
>                 Key: ASTERIXDB-1077
>                 URL: https://issues.apache.org/jira/browse/ASTERIXDB-1077
>             Project: Apache AsterixDB
>          Issue Type: Improvement
>          Components: Hyracks
>            Reporter: Ian Maxon
>            Assignee: Chen Li
>            Priority: Minor
>
> The inverted index tests take up the majority of the time spent in the Hyracks tests
(~25min out of 33min). Taewoo looked into this and discovered that the reason is at least
partially because of the parameters in AccessMethodTestsConfig . Those parameters are not
taking into account that queries on an inverted index will be far more expensive, because
they are all point lookups. It just uses roughly the same number of operations as the BTree/RTree
tests. 
> We may want to lower the number of operations, or perhaps there is some other performance
improvement that may ameliorate this situation. 



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

Mime
View raw message