incubator-blur-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ravikumar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (BLUR-220) Support for humongous Rows
Date Wed, 16 Oct 2013 13:41:43 GMT

    [ https://issues.apache.org/jira/browse/BLUR-220?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13796792#comment-13796792
] 

Ravikumar commented on BLUR-220:
--------------------------------

One thing I want to bring to your attention in the perf-chart

[no.of unique row ids,                       Slow-Query
 total no.of docs]
ID=1M, Docs = 3M                           2721 ms   [Optimized index]
                                                        4486 ms  [Merge-Sorted index, with
early termination]

There is a 2X slowdown for the Slow-Query, which is actually a "common-term" search.

Another variation of the test I did, was to filter-cache the RowId before actually querying.
[Just the bit-set cache of docids]

                                                      Slow-Query
ID=1M, Docs = 3M                           2588 ms   [Optimized index]
                                                         1335 ms  [Merge-Sorted index, with
early termination]

The Merge-Sorted index had a 3.5X jump with this approach and even outperforms an optimized
index

> Support for humongous Rows
> --------------------------
>
>                 Key: BLUR-220
>                 URL: https://issues.apache.org/jira/browse/BLUR-220
>             Project: Apache Blur
>          Issue Type: Improvement
>          Components: Blur
>    Affects Versions: 0.3.0
>            Reporter: Aaron McCurry
>             Fix For: 0.3.0
>
>         Attachments: Blur_Query_Perf_Chart1.pdf, CreateIndex.java, CreateIndex.java,
CreateSortedIndex.java, MyEarlyTerminatingCollector.java, test_results.txt, TestSearch.java,
TestSearch.java
>
>
> One of the limitations of Blur is size of Rows stored, specifically the number of Records.
 The current updates are performed on Lucene is by deleting the document and re-adding to
the index.  Unfortunately when any update is perform on a Row in Blur, the entire Row has
to be re-read (if the RowMutationType is UPDATE_ROW) and then whatever modification needs
are made then it is reindexed in it's entirety.
> Due to all of this overhead, there is a realistic limit on the size of a given Row. 
It may vary based the kind of hardware that is being used, as the Row grows in size the indexing
(mutations) against that Row will slow.
> This issue is being created to discuss techniques on how to deal with this problem.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message