Uh oh!
There was an error while loading. Please reload this page.
[SPARK-29141][SQL][TEST] Use SqlBasedBenchmark in SQL benchmarks - #25828
[SPARK-29141][SQL][TEST] Use SqlBasedBenchmark in SQL benchmarks#25828MaxGekk wants to merge 10 commits into
Conversation
SparkQA
commented
Sep 18, 2019
Test build #110891 has finished for PR 25828 at commit
|
dongjoon-hyun
left a comment
There was a problem hiding this comment.
Thank you so much, @MaxGekk . This looks good.
As a verification, let me regenerate the result on EC2~
dongjoon-hyun
commented
Sep 18, 2019
I updated partially. For the other benchmark test suites like |
dongjoon-hyun
commented
Sep 19, 2019
For |
dongjoon-hyun
left a comment
There was a problem hiding this comment.
+1, LGTM. Merged to master.
This first commit of this PR already pass the Jenkins.
The last two commits are the test result.
SparkQA
commented
Sep 19, 2019
Test build #110951 has finished for PR 25828 at commit
|
SparkQA
commented
Sep 19, 2019
Test build #110947 has finished for PR 25828 at commit
|
| SQL Json 8908 9008 142 1.8 566.42.7X | ||
| SQL Parquet Vectorized 192 229 36 82.112.2125.0X | ||
| SQL Parquet MR 2356 2363 10 6.7149.810.2X | ||
| SQL ORC Vectorized 329 347 25 47.920.9 72.9X |
There was a problem hiding this comment.
ORC Vectorized is almost 2 times slower now. It would be interesting to find the root cause of this.
There was a problem hiding this comment.
Here is the JIRA ticket for that: https://issues.apache.org/jira/browse/SPARK-29169
| Data column - Parquet MR 3378 3384 8 4.7214.811.3X | ||
| Data column - ORC Vectorized 475 481 7 33.130.280.3X | ||
| Data column - ORC MR 2324 2356 46 6.8147.716.4X | ||
| Partition column - CSV 14680 14742 88 1.1 933.32.6X |
There was a problem hiding this comment.
CSV and JSON below is 2 times slower now.
There was a problem hiding this comment.
Here is the JIRA ticket: https://issues.apache.org/jira/browse/SPARK-29170
| SQL CSV 14771 14817 65 0.1 14086.3 1.0X | ||
| SQL Json 29677 29787 157 0.028302.0 0.5X | ||
| SQL Parquet Vectorized 182 191 13 5.8 173.8 81.1X | ||
| SQL Parquet MR 1209 1213 5 0.9 1153.1 12.2X |
There was a problem hiding this comment.
Here is the JIRA ticket: https://issues.apache.org/jira/browse/SPARK-29171
dongjoon-hyun
commented
Sep 19, 2019
Thank you for filing JIRAs. Please add the number directly into that JIRA, too. |
dongjoon-hyun
commented
Sep 19, 2019
For a record, the results were generated based on this PR. So, Scala |
What changes were proposed in this pull request?
Refactored SQL-related benchmark and made them depend on
SqlBasedBenchmark. In particular, creation of Spark session are moved intooverride def getSparkSession: SparkSession.Why are the changes needed?
This should simplify maintenance of SQL-based benchmarks by reducing the number of dependencies. In the future, it should be easier to refactor & extend all SQL benchmarks by changing only one trait. Finally, all SQL-based benchmarks will look uniformly.
Does this PR introduce any user-facing change?
No
How was this patch tested?
By running the modified benchmarks.