hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gopal Vijayaraghavan <gop...@apache.org>
Subject Re: The build-in indexes in ORC file does not work.
Date Wed, 16 Mar 2016 17:45:32 GMT

> I love to see these ORC table optimization help but it is not obvious to
>me under what circumstances they bare fruit.

Are you using Tez or LLAP? Your explain plans are clearly missing the
optimizations I've added as part of Stinger.next.

https://github.com/apache/hive/blob/master/ql/src/test/results/clientpositi
ve/tez/bucketpruning1.q.out#L64


> I am trying to get the whole row for ID = 20. Pretty straight forward
...
>                TableScan
...
> is very misleading as soon as I see it, it gives the impression that it
>is going through the whole table.

Because it *is* a table-scan and the ORC PPD/bloom indexes have
false-positives - a pure index lookup does not need a subsequent Filter.

FYI, if you're looking at simple "select *" with bucketing, you should
vote/watch - https://issues.apache.org/jira/browse/HIVE-12379

Cheers,
Gopal



Mime
View raw message