[opt](serde)Optimize the filling of fixed values into block columns without repeated deserialization. #37377
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
… without repeated deserialization.
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
TPC-H: Total hot run time: 40091 ms |
TPC-DS: Total hot run time: 172692 ms |
ClickBench: Total hot run time: 30.23 s |
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
TPC-H: Total hot run time: 40231 ms |
TPC-DS: Total hot run time: 173854 ms |
ClickBench: Total hot run time: 31 s |
|
run p1 |
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
TPC-H: Total hot run time: 40865 ms |
TPC-DS: Total hot run time: 173943 ms |
ClickBench: Total hot run time: 31.26 s |
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
TPC-H: Total hot run time: 40165 ms |
github-actions
bot
added
the
approved
|
PR approved by at least one committer and no changes requested. |
TPC-DS: Total hot run time: 171645 ms |
ClickBench: Total hot run time: 31.53 s |
AshinGau
approved these changes
Jul 9, 2024
|
LGTM |
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Jul 9, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Jul 10, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Jul 11, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Jul 15, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Jul 15, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
morningman
added
dev/2.1.5-merged
dev/2.1.x
and removed
dev/2.1.x
dev/2.1.5-merged
labels
dataroaring
pushed a commit
that referenced
this pull request
Jul 17, 2024
… without repeated deserialization. (#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
morningman
pushed a commit
that referenced
this pull request
Aug 2, 2024
…rom_fixed_json (#38245) ## Proposed changes fix a bug in DataTypeNullableSerDe.deserialize_column_from_fixed_json. The expected behavior of the `deserialize_column_from_fixed_json` function is to `insert` n values into the column. However, when the `DataTypeNullableSerDe` class implements this function, the null_map column is `resize` to n, which does not insert n values into it. Since this function is only used by the `_fill_partition_columns` of the `parquet/orc reader` and is not called repeatedly for a `get_next_block`, this bug is covered up. before pr : #37377
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Aug 2, 2024
… without repeated deserialization. (apache#37377) ## Proposed changes Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block. ```sql in Hive: CREATE TABLE parquet_partition_tb ( col1 STRING, col2 INT, col3 DOUBLE ) PARTITIONED BY ( partition_col1 STRING, partition_col2 INT ) STORED AS PARQUET; insert into parquet_partition_tb partition (partition_col1="hello",partition_col2=1) values("word",2,2.3); insert into parquet_partition_tb partition(partition_col1="hello",partition_col2=1 ) select col1,col2,col3 from parquet_partition_tb where partition_col1="hello" and partition_col2=1; Repeat the `insert into xxx select xxx`operation several times. Doris : before: mysql> select count(partition_col1) from parquet_partition_tb; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.24 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (3.34 sec) after: mysql> select count(partition_col1) from parquet_partition_tb ; +-----------------------+ | count(partition_col1) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.79 sec) mysql> select count(partition_col2) from parquet_partition_tb; +-----------------------+ | count(partition_col2) | +-----------------------+ | 33554432 | +-----------------------+ 1 row in set (0.51 sec) ``` ## Summary： test sql `select count(partition_col) from tbl;` Number of lines ： 33554432 | |before | after| |---|---|--| |boolean | 3.96|0.47 | |tinyint | 3.39|0.47 | |smallint | 3.14|0.50 | |int |3.34|0.51 | |bigint | 3.61|0.51 | |float | 4.59 |0.51 | |double |4.60| 0.55 | |decimal(5,2)| 3.96 |0.61 | |date | 5.80|0.52 | |timestamp | 7.68 | 0.52 | |string | 3.24 |0.79 | Issue Number: close #xxx <!--Describe your changes.-->
hubgeter
added a commit
to hubgeter/doris
that referenced
this pull request
Aug 2, 2024
…rom_fixed_json (apache#38245) ## Proposed changes fix a bug in DataTypeNullableSerDe.deserialize_column_from_fixed_json. The expected behavior of the `deserialize_column_from_fixed_json` function is to `insert` n values into the column. However, when the `DataTypeNullableSerDe` class implements this function, the null_map column is `resize` to n, which does not insert n values into it. Since this function is only used by the `_fill_partition_columns` of the `parquet/orc reader` and is not called repeatedly for a `get_next_block`, this bug is covered up. before pr : apache#37377
yiguolei
pushed a commit
that referenced
this pull request
Aug 5, 2024
dataroaring
pushed a commit
that referenced
this pull request
Aug 11, 2024
…rom_fixed_json (#38245) ## Proposed changes fix a bug in DataTypeNullableSerDe.deserialize_column_from_fixed_json. The expected behavior of the `deserialize_column_from_fixed_json` function is to `insert` n values into the column. However, when the `DataTypeNullableSerDe` class implements this function, the null_map column is `resize` to n, which does not insert n values into it. Since this function is only used by the `_fill_partition_columns` of the `parquet/orc reader` and is not called repeatedly for a `get_next_block`, this bug is covered up. before pr : #37377
dataroaring
pushed a commit
that referenced
this pull request
Aug 16, 2024
…rom_fixed_json (#38245) ## Proposed changes fix a bug in DataTypeNullableSerDe.deserialize_column_from_fixed_json. The expected behavior of the `deserialize_column_from_fixed_json` function is to `insert` n values into the column. However, when the `DataTypeNullableSerDe` class implements this function, the null_map column is `resize` to n, which does not insert n values into it. Since this function is only used by the `_fill_partition_columns` of the `parquet/orc reader` and is not called repeatedly for a `get_next_block`, this bug is covered up. before pr : #37377
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Proposed changes
Since the value of the partition column is fixed when querying the partition table, we can deserialize the value only once and then repeatedly insert the value into the block.
Summary：
test sql
select count(partition_col) from tbl;Number of lines ： 33554432
Issue Number: close #xxx