Thomas Tauber-Marshall has posted comments on this change.

Change subject: IMPALA-2805: Order conjuncts based on selectivity and cost
......................................................................


Patch Set 13:

I just updated with additional targeted-perf tests. I'm having a little trouble 
getting the new targeted-perf tests to run on the jenkins cluster, but for now 
results from my local machine:

 
+-----------------+---------------------+-----------------------+--------+-------------+------------+-----------+----------------+-------------+-------+
 | Workload        | Query               | File Format           | Avg(s) | 
Base Avg(s) | Delta(Avg) | StdDev(%) | Base StdDev(%) | Num Clients | Iters |
 
+-----------------+---------------------+-----------------------+--------+-------------+------------+-----------+----------------+-------------+-------+
 | TARGETED-PERF() | conjunct_ordering_2 | parquet / none / none | 3.76   | 
4.41        |   -14.74%  |   1.78%   |   0.64%        | 1           | 3     |
 | TARGETED-PERF() | conjunct_ordering_3 | parquet / none / none | 2.11   | 
2.79        | I -24.22%  |   0.13%   |   0.10%        | 1           | 3     |
 | TARGETED-PERF() | conjunct_ordering_4 | parquet / none / none | 0.57   | 
0.77        | I -26.51%  |   0.07%   |   0.20%        | 1           | 3     |
 | TARGETED-PERF() | conjunct_ordering_5 | parquet / none / none | 2.55   | 
4.61        | I -44.58%  |   0.48%   |   0.56%        | 1           | 3     |
 | TARGETED-PERF() | conjunct_ordering_1 | parquet / none / none | 6.60   | 
12.57       | I -47.54%  |   0.78%   |   0.00%        | 1           | 3     |
 
+-----------------+---------------------+-----------------------+--------+-------------+------------+-----------+----------------+-------------+-------+

As for the recent tpcds results, the two queries that look most like possible 
regressions - Q53 and Q96 - were not affected by this change, so their 
performance is probably the result of random variation.

Of the 12 queries - Q7, Q19, Q27, Q34, Q42, Q46, Q52, Q55, Q59, Q63, Q73, Q47 - 
that had their plans changed, only three showed a slow down - Q55, Q46, Q34 - 
and they were all relatively small, so these results look decent.

-- 
To view, visit http://gerrit.cloudera.org:8080/2598
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I02279a26fbc6308ac5eb819d78345fc010469034
Gerrit-PatchSet: 13
Gerrit-Project: Impala
Gerrit-Branch: cdh5-trunk
Gerrit-Owner: Thomas Tauber-Marshall <[email protected]>
Gerrit-Reviewer: Alex Behm <[email protected]>
Gerrit-Reviewer: Henry Robinson <[email protected]>
Gerrit-Reviewer: Marcel Kornacker <[email protected]>
Gerrit-Reviewer: Matthew Jacobs <[email protected]>
Gerrit-Reviewer: Mostafa Mokhtar <[email protected]>
Gerrit-Reviewer: Thomas Tauber-Marshall <[email protected]>
Gerrit-HasComments: No

Reply via email to