Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

add docs for mpp partition table support #7971

Merged
merged 4 commits into from
Mar 29, 2022
Merged

add docs for mpp partition table support #7971

Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
c083849
add docs for mpp partition table support
shichun-0415 Mar 28, 2022
503dcd4
change note to warning
shichun-0415 Mar 28, 2022
59e21d3
Update tiflash/use-tiflash.md
shichun-0415 Mar 28, 2022
f715d86
Update tiflash/use-tiflash.md
shichun-0415 Mar 28, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
65 changes: 65 additions & 0 deletions tiflash/use-tiflash.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -368,6 +368,71 @@ TiFlash provides the following two global/session variables to control whether t
- [`tidb_broadcast_join_threshold_size`](/system-variables.md#tidb_broadcast_join_threshold_count-new-in-v50): The unit of the value is bytes. If the table size (in the unit of bytes) is less than the value of the variable, the Broadcast Hash Join algorithm is used. Otherwise, the Shuffled Hash Join algorithm is used.
- [`tidb_broadcast_join_threshold_count`](/system-variables.md#tidb_broadcast_join_threshold_count-new-in-v50): The unit of the value is rows. If the objects of the join operation belong to a subquery, the optimizer cannot estimate the size of the subquery result set, so the size is determined by the number of rows in the result set. If the estimated number of rows in the subquery is less than the value of this variable, the Broadcast Hash Join algorithm is used. Otherwise, the Shuffled Hash Join algorithm is used.

## Access partitioned tables in MPP mode
shichun-0415 marked this conversation as resolved.
Show resolved Hide resolved

When TiDB accesses partitioned tables in [dynamic pruning mode](https://docs.pingcap.com/tidb/stable/partitioned-table#dynamic-pruning-mode), queries on the partitioned tables are executed in the MPP mode.
shichun-0415 marked this conversation as resolved.
Show resolved Hide resolved

> **Warning:**
>
> Currently, dynamic pruning mode for partitioned tables is still an experimental feature and is not recommended for production environments.

Example:

```sql
mysql> DROP TABLE if exists test.employees;
Query OK, 0 rows affected, 1 warning (0.00 sec)

mysql> CREATE TABLE test.employees ( id int(11) NOT NULL, fname varchar(30) DEFAULT NULL, lname varchar(30) DEFAULT NULL, hired date NOT NULL DEFAULT '1970-01-01', separated date DEFAULT '99
99-12-31', job_code int(11) DEFAULT NULL, store_id int(11) NOT NULL ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin PARTITION BY RANGE (store_id) (PARTITION p0 VALUES LESS THAN (
6), PARTITION p1 VALUES LESS THAN (11), PARTITION p2 VALUES LESS THAN (16), PARTITION p3 VALUES LESS THAN (MAXVALUE));
Query OK, 0 rows affected (0.10 sec)

mysql> ALTER table test.employees SET tiflash replica 1;
Query OK, 0 rows affected (0.09 sec)

mysql> SET tidb_partition_prune_mode=static;
Query OK, 0 rows affected (0.00 sec)

mysql> explain SELECT count(*) FROM test.employees;
+----------------------------------+----------+-------------------+-------------------------------+-----------------------------------+
| id | estRows | task | access object | operator info |
+----------------------------------+----------+-------------------+-------------------------------+-----------------------------------+
| HashAgg_19 | 1.00 | root | | funcs:count(Column#10)->Column#9 |
| └─PartitionUnion_21 | 4.00 | root | | |
| ├─StreamAgg_40 | 1.00 | root | | funcs:count(Column#12)->Column#10 |
| │ └─TableReader_41 | 1.00 | root | | data:StreamAgg_27 |
| │ └─StreamAgg_27 | 1.00 | batchCop[tiflash] | | funcs:count(1)->Column#12 |
| │ └─TableFullScan_39 | 10000.00 | batchCop[tiflash] | table:employees, partition:p0 | keep order:false, stats:pseudo |
| ├─StreamAgg_63 | 1.00 | root | | funcs:count(Column#14)->Column#10 |
| │ └─TableReader_64 | 1.00 | root | | data:StreamAgg_50 |
| │ └─StreamAgg_50 | 1.00 | batchCop[tiflash] | | funcs:count(1)->Column#14 |
| │ └─TableFullScan_62 | 10000.00 | batchCop[tiflash] | table:employees, partition:p1 | keep order:false, stats:pseudo |
| ├─StreamAgg_86 | 1.00 | root | | funcs:count(Column#16)->Column#10 |
| │ └─TableReader_87 | 1.00 | root | | data:StreamAgg_73 |
| │ └─StreamAgg_73 | 1.00 | batchCop[tiflash] | | funcs:count(1)->Column#16 |
| │ └─TableFullScan_85 | 10000.00 | batchCop[tiflash] | table:employees, partition:p2 | keep order:false, stats:pseudo |
| └─StreamAgg_109 | 1.00 | root | | funcs:count(Column#18)->Column#10 |
| └─TableReader_110 | 1.00 | root | | data:StreamAgg_96 |
| └─StreamAgg_96 | 1.00 | batchCop[tiflash] | | funcs:count(1)->Column#18 |
| └─TableFullScan_108 | 10000.00 | batchCop[tiflash] | table:employees, partition:p3 | keep order:false, stats:pseudo |
+----------------------------------+----------+-------------------+-------------------------------+-----------------------------------+
18 rows in set, 4 warnings (0.00 sec)

mysql> SET tidb_partition_prune_mode=dynamic;
Query OK, 0 rows affected (0.00 sec)

mysql> explain SELECT count(*) FROM test.employees;
+------------------------------+----------+-------------------+-----------------+----------------------------------+
| id | estRows | task | access object | operator info |
+------------------------------+----------+-------------------+-----------------+----------------------------------+
| HashAgg_21 | 1.00 | root | | funcs:count(Column#11)->Column#9 |
| └─TableReader_23 | 1.00 | root | partition:all | data:ExchangeSender_22 |
| └─ExchangeSender_22 | 1.00 | batchCop[tiflash] | | ExchangeType: PassThrough |
| └─HashAgg_9 | 1.00 | batchCop[tiflash] | | funcs:count(1)->Column#11 |
| └─TableFullScan_20 | 10000.00 | batchCop[tiflash] | table:employees | keep order:false, stats:pseudo |
+------------------------------+----------+-------------------+-----------------+----------------------------------+
```

## Data validation

### User scenarios
Expand Down