[
https://issues.apache.org/jira/browse/IMPALA-7560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16619827#comment-16619827
]
Paul Rogers commented on IMPALA-7560:
-------------------------------------
Turns out that Apache Drill did a similar analysis to work out rules based on
the classic defaults plus some reasoning about probability: DRILL-5254
For Drill, since only the "classic" estimates (not stats) are available, the
probabilities don't work out because of he conditional probability of a user
using one operator vs. another. But, the math reasoning might be used here if
we do have stats to work with.
> Better selectivity estimate for != (not equals) binary predicate
> ----------------------------------------------------------------
>
> Key: IMPALA-7560
> URL: https://issues.apache.org/jira/browse/IMPALA-7560
> Project: IMPALA
> Issue Type: Bug
> Components: Frontend
> Affects Versions: Impala 2.8.0, Impala 2.9.0, Impala 2.10.0, Impala
> 2.12.0, Impala 2.13.0
> Reporter: bharath v
> Priority: Major
>
> Currently we use the default selectivity estimate for any binary predicate
> with op other than EQ / NON_DISTINCT.
> {noformat}
> // Determine selectivity
> // TODO: Compute selectivity for nested predicates.
> // TODO: Improve estimation using histograms.
> Reference<SlotRef> slotRefRef = new Reference<SlotRef>();
> if ((op_ == Operator.EQ || op_ == Operator.NOT_DISTINCT)
> && isSingleColumnPredicate(slotRefRef, null)) {
> long distinctValues = slotRefRef.getRef().getNumDistinctValues();
> if (distinctValues > 0) {
> selectivity_ = 1.0 / distinctValues;
> selectivity_ = Math.max(0, Math.min(1, selectivity_));
> }
> }
> {noformat}
> This can give very conservative estimates. For example:
> {noformat}
> [localhost:21000] tpch> select * from nation where n_regionkey != 1;
> [localhost:21000] tpch> summary;
> +--------------+--------+----------+----------+-------+------------+-----------+---------------+-------------+
> | Operator | #Hosts | Avg Time | Max Time | *#Rows* | *Est. #Rows* | Peak
> Mem | Est. Peak Mem | Detail |
> +--------------+--------+----------+----------+-------+------------+-----------+---------------+-------------+
> | 00:SCAN HDFS | 1 | 3.32ms | 3.32ms | *20* | *3* |
> 143.00 KB | 16.00 MB | tpch.nation |
> +--------------+--------+----------+----------+-------+------------+-----------+---------------+-------------+
> [localhost:21000] tpch>
> {noformat}
> Ideally we could've inversed the selecitivity to 4/5 (=1 - 1/5) that can
> give better estimate.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org
For additional commands, e-mail: issues-all-h...@impala.apache.org
