planner: support expand IN expr

[Yriuns]

What problem does this PR solve?

Issue Number: close #34882

Problem Summary:

Provide a hint that allows optimizer to expand the in expression.

A selection like a = 0 AND b = 1 AND c IN (2, 3) will be convert to something like this:

a = 0 AND b = 1 AND c = 2
UNION ALL
a = 0 AND b = 1 AND c = 3

So that the we can utilize the index better, e.g., TopN push down can be optimized as LimitN.

What is changed and how it works?

For WHERE clause with IN expressions, we divide the expressions list into non-IN expressions list and IN expressions list. For the lists of IN expressions, we do a Cartesian product. Each element of the Cartesian product result set is merged with the lists of non-IN expressions to obtain a new WHERE clause. Finally, we use a UNION ALL to merge all these new Selection operators.

An example:

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)

mysql> CREATE TABLE `t` (
    ->   `id` bigint(22) NOT NULL AUTO_INCREMENT,
    ->   `a` bigint(20) NOT NULL DEFAULT '0',
    ->   `b` bigint(20) NOT NULL DEFAULT '0',
    ->   `c` bigint(20) NOT NULL DEFAULT '0',
    ->   `d` bigint(20) NOT NULL DEFAULT '0',
    ->   UNIQUE KEY `uk_id` (`id`),
    ->   KEY `a_b_c_id` (`a`,`b`,`c`,`id`)
    -> ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin AUTO_INCREMENT=1;

-- without hint
mysql> explain SELECT id, a, b, c, d FROM t WHERE a = 0 AND b = 1 AND c IN (2, 3) ORDER BY id ASC LIMIT 10;
+----------------------------------+---------+-----------+--------------------------------------+--------------------------------------------------------------------+
| id                               | estRows | task      | access object                        | operator info                                                      |
+----------------------------------+---------+-----------+--------------------------------------+--------------------------------------------------------------------+
| TopN_9                           | 0.00    | root      |                                      | test.t.id, offset:0, count:10                                      |
| └─IndexLookUp_20                 | 0.00    | root      |                                      |                                                                    |
|   ├─TopN_19(Build)               | 0.00    | cop[tikv] |                                      | test.t.id, offset:0, count:10                                      |
|   │ └─IndexRangeScan_17          | 0.00    | cop[tikv] | table:t, index:a_b_c_id(a, b, c, id) | range:[0 1 2,0 1 2], [0 1 3,0 1 3], keep order:false, stats:pseudo |
|   └─TableRowIDScan_18(Probe)     | 0.00    | cop[tikv] | table:t                              | keep order:false, stats:pseudo                                     |
+----------------------------------+---------+-----------+--------------------------------------+--------------------------------------------------------------------+
5 rows in set (0.00 sec)

-- with hint
mysql> explain SELECT /*+ in_expansion() */ id, a, b, c, d FROM t WHERE a = 0 AND b = 1 AND c IN (2, 3) ORDER BY id ASC LIMIT 10;
+------------------------------------------+---------+-----------+--------------------------------------+----------------------------------------------------+
| id                                       | estRows | task      | access object                        | operator info                                      |
+------------------------------------------+---------+-----------+--------------------------------------+----------------------------------------------------+
| TopN_17                                  | 0.00    | root      |                                      | test.t.id, offset:0, count:10                      |
| └─Union_22                               | 0.00    | root      |                                      |                                                    |
|   ├─Limit_27                             | 0.00    | root      |                                      | offset:0, count:10                                 |
|   │ └─Projection_40                      | 0.00    | root      |                                      | test.t.id, test.t.a, test.t.b, test.t.c, test.t.d  |
|   │   └─IndexLookUp_39                   | 0.00    | root      |                                      |                                                    |
|   │     ├─Limit_38(Build)                | 0.00    | cop[tikv] |                                      | offset:0, count:10                                 |
|   │     │ └─IndexRangeScan_36            | 0.00    | cop[tikv] | table:t, index:a_b_c_id(a, b, c, id) | range:[0 1 2,0 1 2], keep order:true, stats:pseudo |
|   │     └─TableRowIDScan_37(Probe)       | 0.00    | cop[tikv] | table:t                              | keep order:false, stats:pseudo                     |
|   └─Limit_45                             | 0.00    | root      |                                      | offset:0, count:10                                 |
|     └─Projection_58                      | 0.00    | root      |                                      | test.t.id, test.t.a, test.t.b, test.t.c, test.t.d  |
|       └─IndexLookUp_57                   | 0.00    | root      |                                      |                                                    |
|         ├─Limit_56(Build)                | 0.00    | cop[tikv] |                                      | offset:0, count:10                                 |
|         │ └─IndexRangeScan_54            | 0.00    | cop[tikv] | table:t, index:a_b_c_id(a, b, c, id) | range:[0 1 3,0 1 3], keep order:true, stats:pseudo |
|         └─TableRowIDScan_55(Probe)       | 0.00    | cop[tikv] | table:t                              | keep order:false, stats:pseudo                     |
+------------------------------------------+---------+-----------+--------------------------------------+----------------------------------------------------+
14 rows in set (0.00 sec)

• No code

As we can see, after we use IN_EXPANSION() hint, the TopN_19 to tikv converts to 2 Limit to tikv. with-hint plan only needs to scan at most 10 + 10 records, but without-hint plan needs to scan and sort all the records of ranges [0 1 2,0 1 2], [0 1 3,0 1 3]. If there are a lot of records of these ranges, we can save a lot of uncessary cost.

I also conduct a micro benchmark(1 concurrency of client) in my own server. Here is the result:

CREATE TABLE `t` (
  `id` bigint(22) NOT NULL AUTO_INCREMENT,
  `a` bigint(22) NOT NULL,
  `b` bigint(22) NOT NULL,
  `c` bigint(22) NOT NULL,
  `d` bigint(22) NOT NULL,
  KEY `a_b_c_d_id` (`a`,`b`,`c`,`d`, `id`),
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin AUTO_INCREMENT=

SELECT *
FROM t
WHERE a = 14 AND b IN (3, 4, 7, 8, 11, 12, 13) AND c = 1 AND d IN (0, 20)
ORDER BY id asc
LIMIT 0, 20

Without hint

number of records that satisfy filter condition	QPS	avg latency	TP99	TiDB CPU
1000	277.01	3.6	7.17	150%
10000	65.23	15.33	26.58	30%
100000	6.28	159.32	170.48	<10%

With hint

number of records that satisfy filter condition	QPS	avg latency	TP99	TiDB CPU
1000	140.45	7.12	11.87	740%
10000	136.05	7.35	12.30	740%
100000	134.05	7.46	12.30	750%

The latency is quite stable, and the more records, the more time we save.

Side effects

• Performance regression: Consumes more CPU of tidb, but save CPU of tikv
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

None

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has not been approved.

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[sre-bot]

All committers have signed the CLA.

[ti-chi-bot]

Welcome @Yriuns!

It looks like this is your first PR to pingcap/tidb 🎉.

I'm the bot to help you request reviewers, add labels and more, See available commands.

We want to make sure your contribution gets all the attention it needs!



Thank you, and welcome to pingcap/tidb. 😃

[Yriuns]

/cc @winoros

[Yriuns]

/cc @time-and-fate

[ti-chi-bot]

@Yriuns: PR needs rebase.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.