[GitHub] carbondata pull request #2567: [CARBONDATA-2790][BloomDataMap]Optimize defau...

[qiuchenjian-2]

GitHub user xuchuanyin opened a pull request:

    https://github.com/apache/carbondata/pull/2567

    [CARBONDATA-2790][BloomDataMap]Optimize default parameter for bloomfilter datamap

    To provide better query performance for bloomfilter datamap by default,
    we optimize bloom_size from 32000 to 640000 and optimize bloom_fpp from
    0.01 to 0.00001.
   
    Be sure to do all of the following checklist to help us incorporate
    your contribution quickly and easily:
   
     - [ ] Any interfaces changed?
     
     - [ ] Any backward compatibility impacted?
     
     - [ ] Document update required?
   
     - [ ] Testing done
            Please provide details on
            - Whether new unit test cases have been added or why no new tests are required?
            - How it is tested? Please attach test report.
            - Is it a performance related change? Please attach the performance test report.
            - Any additional information to help reviewers in testing this change.
           
     - [ ] For large changes, please consider breaking it into sub-tasks under an umbrella JIRA.
   


You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/xuchuanyin/carbondata 0727_bloom_default_para

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/carbondata/pull/2567.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #2567
   
----
commit 50780e8159ae46e6a7c9596ee9d1ad8dd09048a5
Author: xuchuanyin <xuchuanyin@...>
Date:   2018-07-27T03:54:21Z

    Optimize default parameter for bloomfilter datamap
   
    To provide better query performance for bloomfilter datamap by default,
    we optimize bloom_size from 32000 to 640000 and optimize bloom_fpp from
    0.01 to 0.00001.

----


---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Failed  with Spark 2.1.0, Please check CI http://136.243.101.176:8080/job/ApacheCarbonPRBuilder1/7547/



---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Failed with Spark 2.2.1, Please check CI http://88.99.58.216:8080/job/ApacheCarbonPRBuilder/6300/



---

[qiuchenjian-2]

Github user ravipesala commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    SDV Build Fail , Please check CI http://144.76.159.231:8080/job/ApacheSDVTests/6023/



---

[qiuchenjian-2]

Github user brijoobopanna commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    retest this please


---

[qiuchenjian-2]

Github user brijoobopanna commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    retest sdv please


---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Success with Spark 2.1.0, Please check CI http://136.243.101.176:8080/job/ApacheCarbonPRBuilder1/7589/



---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Failed with Spark 2.2.1, Please check CI http://88.99.58.216:8080/job/ApacheCarbonPRBuilder/6344/



---

[qiuchenjian-2]

Github user ravipesala commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    SDV Build Fail , Please check CI http://144.76.159.231:8080/job/ApacheSDVTests/6052/



---

[qiuchenjian-2]

Github user xuchuanyin commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    retest this please


---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Failed  with Spark 2.1.0, Please check CI http://136.243.101.176:8080/job/ApacheCarbonPRBuilder1/7648/



---

[qiuchenjian-2]

Github user jackylk commented on a diff in the pull request:

    https://github.com/apache/carbondata/pull/2567#discussion_r206416298
 
    --- Diff: docs/datamap/bloomfilter-datamap-guide.md ---
    @@ -83,8 +83,8 @@ User can create BloomFilter datamap using the Create DataMap DDL:
     | Property | Is Required | Default Value | Description |
     |-------------|----------|--------|---------|
     | INDEX_COLUMNS | YES |  | Carbondata will generate BloomFilter index on these columns. Queries on there columns are usually like 'COL = VAL'. |
    -| BLOOM_SIZE | NO | 32000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    -| BLOOM_FPP | NO | 0.01 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    +| BLOOM_SIZE | NO | 640000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    --- End diff --
   
    please modify `the value 32000 * #noOfPagesInBlocklet`


---

[qiuchenjian-2]

Github user jackylk commented on a diff in the pull request:

    https://github.com/apache/carbondata/pull/2567#discussion_r206416739
 
    --- Diff: docs/datamap/bloomfilter-datamap-guide.md ---
    @@ -83,8 +83,8 @@ User can create BloomFilter datamap using the Create DataMap DDL:
     | Property | Is Required | Default Value | Description |
     |-------------|----------|--------|---------|
     | INDEX_COLUMNS | YES |  | Carbondata will generate BloomFilter index on these columns. Queries on there columns are usually like 'COL = VAL'. |
    -| BLOOM_SIZE | NO | 32000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    -| BLOOM_FPP | NO | 0.01 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    +| BLOOM_SIZE | NO | 640000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    --- End diff --
   
    please modify `so the value is the approximate records in a blocklet` to `so the default value is the approximate records in a blocklet assuming the blocklet size is ???`


---

[qiuchenjian-2]

Github user jackylk commented on a diff in the pull request:

    https://github.com/apache/carbondata/pull/2567#discussion_r206417322
 
    --- Diff: docs/datamap/bloomfilter-datamap-guide.md ---
    @@ -83,8 +83,8 @@ User can create BloomFilter datamap using the Create DataMap DDL:
     | Property | Is Required | Default Value | Description |
     |-------------|----------|--------|---------|
     | INDEX_COLUMNS | YES |  | Carbondata will generate BloomFilter index on these columns. Queries on there columns are usually like 'COL = VAL'. |
    -| BLOOM_SIZE | NO | 32000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    -| BLOOM_FPP | NO | 0.01 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    +| BLOOM_SIZE | NO | 640000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    +| BLOOM_FPP | NO | 0.00001 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    --- End diff --
   
    explain `the default value is set to 0.00001 to make low false-positive while keep bloomfilter in acceptable size`. Or can you provide some data for the size?


---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Success with Spark 2.2.1, Please check CI http://88.99.58.216:8080/job/ApacheCarbonPRBuilder/6381/



---

[qiuchenjian-2]

Github user xuchuanyin commented on a diff in the pull request:

    https://github.com/apache/carbondata/pull/2567#discussion_r206459733
 
    --- Diff: docs/datamap/bloomfilter-datamap-guide.md ---
    @@ -83,8 +83,8 @@ User can create BloomFilter datamap using the Create DataMap DDL:
     | Property | Is Required | Default Value | Description |
     |-------------|----------|--------|---------|
     | INDEX_COLUMNS | YES |  | Carbondata will generate BloomFilter index on these columns. Queries on there columns are usually like 'COL = VAL'. |
    -| BLOOM_SIZE | NO | 32000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    -| BLOOM_FPP | NO | 0.01 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    +| BLOOM_SIZE | NO | 640000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    --- End diff --
   
    OK， changed it to
    ```
    so the default value is the approximate distinct index values in a blocklet assuming the blocklet size is 20 pages and each page contains 32000 records.
    ```


---

[qiuchenjian-2]

Github user xuchuanyin commented on a diff in the pull request:

    https://github.com/apache/carbondata/pull/2567#discussion_r206462541
 
    --- Diff: docs/datamap/bloomfilter-datamap-guide.md ---
    @@ -83,8 +83,8 @@ User can create BloomFilter datamap using the Create DataMap DDL:
     | Property | Is Required | Default Value | Description |
     |-------------|----------|--------|---------|
     | INDEX_COLUMNS | YES |  | Carbondata will generate BloomFilter index on these columns. Queries on there columns are usually like 'COL = VAL'. |
    -| BLOOM_SIZE | NO | 32000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    -| BLOOM_FPP | NO | 0.01 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    +| BLOOM_SIZE | NO | 640000 | This value is internally used by BloomFilter as the number of expected insertions, it will affects the size of BloomFilter index. Since each blocklet has a BloomFilter here, so the value is the approximate records in a blocklet. In another word, the value 32000 * #noOfPagesInBlocklet. The value should be an integer. |
    +| BLOOM_FPP | NO | 0.00001 | This value is internally used by BloomFilter as the False-Positive Probability, it will affects the size of bloomfilter index as well as the number of hash functions for the BloomFilter. The value should be in range (0, 1). |
    --- End diff --
   
    Added example
    ```
    In one test scenario, a 96GB TPCH consumer table with bloom_size=320000 and bloom_fpp=0.00001 will result in 18 false positive samples
    ```


---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Failed  with Spark 2.1.0, Please check CI http://136.243.101.176:8080/job/ApacheCarbonPRBuilder1/7674/



---

[qiuchenjian-2]

Github user CarbonDataQA commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    Build Success with Spark 2.2.1, Please check CI http://88.99.58.216:8080/job/ApacheCarbonPRBuilder/6406/



---

[qiuchenjian-2]

Github user brijoobopanna commented on the issue:

    https://github.com/apache/carbondata/pull/2567
 
    retest this please



---