executor: add coprocessor cache hit ratio in explain analyze

[lzmhhh123]

What problem does this PR solve?

Issue Number: close #17370

What is changed and how it works?

How it Works: add coprocessor cache hit rate in explain analyze.

here is an example:

tidb(localhost:4001) > explain analyze select * from t use index(a);
+-------------------------------+-----------+---------+-----------+------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+-----------------------+------+
| id                            | estRows   | actRows | task      | access object          | execution info                                                                                                                                                                                                                                           | operator info                  | memory                | disk |
+-------------------------------+-----------+---------+-----------+------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+-----------------------+------+
| IndexLookUp_6                 | 262400.00 | 262400  | root      |                        | time:620.513742ms, loops:258, cop_task: {num: 4, max: 5.530817ms, min: 1.51829ms, avg: 2.70883ms, p95: 5.530817ms, max_proc_keys: 2480, p95_proc_keys: 2480, tot_proc: 1ms, tot_wait: 1ms, rpc_num: 4, rpc_time: 10.816328ms, copr_cache_hit_rate: 0.75} |                                | 6.685169219970703 MB  | N/A  |
| ├─IndexFullScan_4(Build)      | 262400.00 | 262400  | cop[tikv] | table:t, index:a(a, c) | proc max:93ms, min:1ms, p80:93ms, p95:93ms, iters:275, tasks:4                                                                                                                                                                                           | keep order:false, stats:pseudo | 1.7549400329589844 MB | N/A  |
| └─TableRowIDScan_5(Probe)     | 262400.00 | 0       | cop[tikv] | table:t                | time:0ns, loops:0                                                                                                                                                                                                                                        | keep order:false, stats:pseudo | N/A                   | N/A  |
+-------------------------------+-----------+---------+-----------+------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+-----------------------+------+
3 rows in set (0.62 sec)

another example with consecutive two same queries.

tidb(localhost:4001) > explain analyze select * from t;
+-----------------------+-----------+---------+-----------+---------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------+--------------------------------+----------------------+------+
| id                    | estRows   | actRows | task      | access object | execution info
                                                                     | operator info                  | memory               | disk |
+-----------------------+-----------+---------+-----------+---------------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+----------------------+------+
| TableReader_5         | 262400.00 | 262400  | root      |               | time:398.505526ms, loops:258, cop_task: {num: 3, max: 395.43793ms, min: 126.256914ms, avg: 284.324103ms, p95: 395.43793ms, max_proc_keys: 103120, p95_proc_keys: 103120, tot_proc: 421ms, tot_wait: 11ms, rpc_num: 3, rpc_time: 852.957604ms, copr_cache_hit_rate: 0.00} | data:TableFullScan_4           | 10.83200740814209 MB | N/A  |
| └─TableFullScan_4     | 262400.00 | 262400  | cop[tikv] | table:t       | proc max:159ms, min:41ms, p80:159ms, p95:159ms, iters:269, tasks:3                                                                                                                                                                                                       | keep order:false, stats:pseudo | N/A                  | N/A  |
+-----------------------+-----------+---------+-----------+---------------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+----------------------+------+
2 rows in set (0.42 sec)

tidb(localhost:4001) > explain analyze select * from t;
+-----------------------+-----------+---------+-----------+---------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+-----------------------+------+
| id                    | estRows   | actRows | task      | access object | execution info                                                                                                                                                                | operator info                  | memory                | disk |
+-----------------------+-----------+---------+-----------+---------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------------------------------+-----------------------+------+
| TableReader_5         | 262400.00 | 262400  | root      |               | time:8.316905ms, loops:258, cop_task: {num: 3, max: 2.68057ms, min: 2.104203ms, avg: 2.332752ms, p95: 2.68057ms, rpc_num: 3, rpc_time: 6.982763ms, copr_cache_hit_rate: 1.00} | data:TableFullScan_4
           | 10.831978797912598 MB | N/A  |
| └─TableFullScan_4     | 262400.00 | 262400  | cop[tikv] | table:t       | proc max:159ms, min:41ms, p80:159ms, p95:159ms, iters:269, tasks:3                                                                                                            | keep order:false, st
ats:pseudo | N/A                   | N/A  |
+-----------------------+-----------+---------+-----------+---------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------
-----------+-----------------------+------+
2 rows in set (0.02 sec)

Related changes

• Need to cherry-pick to the release branch

Check List

Tests

• Unit test
• Integration test

Side effects

• Performance regression
  • Consumes more CPU

Release note

• add coprocessor cache hit ratio in explain analyze.

[qw4990]

LGTM

[ghost]

I would call this a hit ratio, rather than a rate. These are often used interchangeably, but I try to be careful to use rate to refer to the number of misses over a time scale. i.e. 123 misses/s

Edit: Google seems to back me on this one. A rate is for comparing different units (such as cache misses over time):

A ratio is a comparison of two numbers or measurements. The numbers or measurements being compared are called the terms of the ratio. A rate is a special ratio in which the two terms are in different units.

[lzmhhh123]

I would call this a hit ratio, rather than a rate. These are often used interchangeably, but I try to be careful to use rate to refer to the number of misses over a time scale. i.e. 123 misses/s

Edit: Google seems to back me on this one. A rate is for comparing different units (such as cache misses over time):

A ratio is a comparison of two numbers or measurements. The numbers or measurements being compared are called the terms of the ratio. A rate is a special ratio in which the two terms are in different units.

Nice catch.

[ti-srebot]

@nullnotnil,Thanks for your review. The bot only counts LGTMs from Reviewers and higher roles, but you're still welcome to leave your comments.See the corresponding SIG page for more information. Related SIG: execution(slack).

Clear review, I meant to just dismiss previous (LGTM but I'm not a LGTM'er)

[crazycs520]

LGTM

[lzmhhh123]

/merge

[ti-srebot]

/run-all-tests

[ti-srebot]

cherry pick to release-4.0 in PR #19972