[20190524]使用use_concat or_expand提示优化.txt

[20190524]使用use_concat or_expand提示优化.txt

–//上午看了链接https://connor-mcdonald.com/2019/05/22/being-generous-to-the-optimizer,突然想起我们生产系统类似语句。
–//现在想想觉得开发的想象力太丰富,写这些语句是否考虑长期运行导致的结果。对方例子相对简单,我优化的例子简直就是变态。
–//链接:[20150814]使用use_concat提示.txt => http://blog.itpub.net/267265/viewspace-1771727/
–//实际上看了马上想到使用use_concat or_expand提示优化sql语句.同时看了链接
–//https://jonathanlewis.wordpress.com/2019/05/22/danger-hints/,一起测试看看。

1.环境:
scott@test01p> @ ver1
port_string                    version        banner                                                                               con_id
—————————— ————– ——————————————————————————– ———-
ibmpc/win_nt64-9.1.0           12.2.0.1.0     oracle database 12c enterprise edition release 12.2.0.1.0 – 64bit production              0

create table address ( street int, suburb int, post_code int,  data char(100));
insert into address select mod(rownum,1e4), mod(rownum,10), mod(rownum,1e2), rownum from dual connect by level  <= 1e5;
commit;
 
exec dbms_stats.gather_table_stats(”,’address’)

create index i_address_stress on address ( street );
create index i_address_suburb on address ( suburb );
create index i_address_post_code on address ( post_code );

2.测试:

variable val number = 6
variable choice number = 1
alter session set statistics_level = all;

scott@test01p> select data from   address where ( :choice = 1 and street = :val ) or ( :choice = 2 and suburb = :val );
data
—–
6
10006
20006
30006
40006
50006
60006
70006
80006
90006
10 rows selected.

plan hash value: 3645838471
———————————————————————————————————————–
| id  | operation         | name    | starts | e-rows |e-bytes| cost (%cpu)| e-time   | a-rows |   a-time   | buffers |
———————————————————————————————————————–
|   0 | select statement  |         |      1 |        |       |   445 (100)|          |     10 |00:00:00.02 |    1636 |
|*  1 |  table access full| address |      1 |    100 | 10800 |   445   (1)| 00:00:01 |     10 |00:00:00.02 |    1636 |
———————————————————————————————————————–
query block name / object alias (identified by operation id):
————————————————————-
   1 – sel$1 / address@sel$1
peeked binds (identified by position):
————————————–
   2 – :2 (number): 6
   4 – :2 (number, primary=2)
predicate information (identified by operation id):
—————————————————
   1 – filter(((:choice=2 and “suburb”=:val) or (“street”=:val and :choice=1)))

–//选择全表扫描.加入提示:/*+ or_expand(@sel$1) */

select /*+ or_expand(@sel$1) */ data from   address where ( :choice = 1 and street = :val ) or ( :choice = 2 and suburb = :val );

scott@test01p> @ dpc ” outline
plan hash value: 1427591975
————————————————————————————————————————————————————–
| id  | operation                              | name             | starts | e-rows |e-bytes| cost (%cpu)| e-time   | a-rows |   a-time   | buffers | reads  |
————————————————————————————————————————————————————–
|   0 | select statement                       |                  |      1 |        |       |   456 (100)|          |     10 |00:00:00.06 |      13 |      4 |
|   1 |  view                                  | vw_ore_b7380f92  |      1 |  10010 |   997k|   456   (1)| 00:00:01 |     10 |00:00:00.06 |      13 |      4 |
|   2 |   union-all                            |                  |      1 |        |       |            |          |     10 |00:00:00.06 |      13 |      4 |
|*  3 |    filter                              |                  |      1 |        |       |            |          |     10 |00:00:00.06 |      13 |      4 |
|   4 |     table access by index rowid batched| address          |      1 |     10 |  1050 |    11   (0)| 00:00:01 |     10 |00:00:00.06 |      13 |      4 |
|*  5 |      index range scan                  | i_address_stress |      1 |     10 |       |     1   (0)| 00:00:01 |     10 |00:00:00.06 |       3 |      4 |
|*  6 |    filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |      0 |
|*  7 |     table access full                  | address          |      0 |  10000 |  1054k|   445   (1)| 00:00:01 |      0 |00:00:00.01 |       0 |      0 |
————————————————————————————————————————————————————–
query block name / object alias (identified by operation id):
————————————————————-
   1 – set$9162bf3c   / vw_ore_b7380f92@sel$b7380f92
   2 – set$9162bf3c
   3 – set$9162bf3c_1
   4 – set$9162bf3c_1 / address@sel$1
   5 – set$9162bf3c_1 / address@sel$1
   6 – set$9162bf3c_2
   7 – set$9162bf3c_2 / address@sel$1
outline data
————-
  /*+
      begin_outline_data
      ignore_optim_embedded_hints
      optimizer_features_enable(‘12.2.0.1’)
      db_version(‘12.2.0.1’)
      all_rows
      outline_leaf(@”set$9162bf3c_2″)
      outline_leaf(@”set$9162bf3c_1″)
      outline_leaf(@”set$9162bf3c”)
      or_expand(@”sel$1″ (1) (2))
      outline_leaf(@”sel$b7380f92″)
      outline(@”set$9162bf3c”)
      or_expand(@”sel$1″ (1) (2))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~      
      outline(@”sel$1″)
      no_access(@”sel$b7380f92″ “vw_ore_b7380f92″@”sel$b7380f92”)
      index_rs_asc(@”set$9162bf3c_1″ “address”@”sel$1” (“address”.”street”))
      batch_table_access_by_rowid(@”set$9162bf3c_1″ “address”@”sel$1”)
      full(@”set$9162bf3c_2″ “address”@”sel$1”)
      end_outline_data
  */
peeked binds (identified by position):
————————————–
   2 – :2 (number): 6
   4 – :2 (number, primary=2)
predicate information (identified by operation id):
—————————————————
   3 – filter(:choice=1)
   5 – access(“street”=:val)
   6 – filter(:choice=2)
   7 – filter((“suburb”=:val and (lnnvl(:choice=1) or lnnvl(“street”=:val))))
–//注意看下划线.
–//但是使用下划线提示,改动代码的情况下如何呢?

select /*+ or_expand(@sel$1 (1) (2) ) */ data
  from address
 where ( :choice = 1 and street = :val )
    or ( :choice = 2 and suburb = :val )
    or ( :choice = 3 and post_code = :val);

scott@test01p> @ dpc ” outline

plan hash value: 1427591975
—————————————————————————————————————————————————–
| id  | operation                              | name             | starts | e-rows |e-bytes| cost (%cpu)| e-time   | a-rows |   a-time   | buffers |
—————————————————————————————————————————————————–
|   0 | select statement                       |                  |      1 |        |       |   456 (100)|          |     10 |00:00:00.01 |      13 |
|   1 |  view                                  | vw_ore_b7380f92  |      1 |  10010 |   997k|   456   (1)| 00:00:01 |     10 |00:00:00.01 |      13 |
|   2 |   union-all                            |                  |      1 |        |       |            |          |     10 |00:00:00.01 |      13 |
|*  3 |    filter                              |                  |      1 |        |       |            |          |     10 |00:00:00.01 |      13 |
|   4 |     table access by index rowid batched| address          |      1 |     10 |  1050 |    11   (0)| 00:00:01 |     10 |00:00:00.01 |      13 |
|*  5 |      index range scan                  | i_address_stress |      1 |     10 |       |     1   (0)| 00:00:01 |     10 |00:00:00.01 |       3 |
|*  6 |    filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |
|*  7 |     table access full                  | address          |      0 |  10000 |  1054k|   445   (1)| 00:00:01 |      0 |00:00:00.01 |       0 |
—————————————————————————————————————————————————–
query block name / object alias (identified by operation id):
————————————————————-

   1 – set$9162bf3c   / vw_ore_b7380f92@sel$b7380f92
   2 – set$9162bf3c
   3 – set$9162bf3c_1
   4 – set$9162bf3c_1 / address@sel$1
   5 – set$9162bf3c_1 / address@sel$1
   6 – set$9162bf3c_2
   7 – set$9162bf3c_2 / address@sel$1

outline data
————-

  /*+
      begin_outline_data
      ignore_optim_embedded_hints
      optimizer_features_enable(‘12.2.0.1’)
      db_version(‘12.2.0.1’)
      all_rows
      outline_leaf(@”set$9162bf3c_2″)
      outline_leaf(@”set$9162bf3c_1″)
      outline_leaf(@”set$9162bf3c”)
      or_expand(@”sel$1″ (1) (2))
      outline_leaf(@”sel$b7380f92″)
      outline(@”set$9162bf3c”)
      or_expand(@”sel$1″ (1) (2))
      outline(@”sel$1″)
      no_access(@”sel$b7380f92″ “vw_ore_b7380f92″@”sel$b7380f92”)
      index_rs_asc(@”set$9162bf3c_1″ “address”@”sel$1” (“address”.”street”))
      batch_table_access_by_rowid(@”set$9162bf3c_1″ “address”@”sel$1”)
      full(@”set$9162bf3c_2″ “address”@”sel$1”)
      end_outline_data
  */

peeked binds (identified by position):
————————————–

   2 – :2 (number): 6
   4 – :2 (number, primary=2)

predicate information (identified by operation id):
—————————————————

   3 – filter(:choice=1)
   5 – access(“street”=:val)
   6 – filter(:choice=2)
   7 – filter((“suburb”=:val and (lnnvl(:choice=1) or lnnvl(“street”=:val))))

–//正像链接讲的那样如果增加1个或条件,导致执行计划变得不合理.实际上跟严重的是查询发生了错误.
–//如果仔细看predicate information 就很容易发现没有:choice=1的filter.如果查询:

scott@test01p> variable choice number = 3
select /*+ or_expand(@sel$1 (1) (2) ) */ data
  from address
 where ( :choice = 1 and street = :val )
    or ( :choice = 2 and suburb = :val )
    or ( :choice = 3 and post_code = :val);

no rows selected.

–//取消提示:
select data  from address where ( :choice = 1 and street = :val ) or ( :choice = 2 and suburb = :val ) or ( :choice = 3 and post_code = :val);

1000 rows selected.

–//两者的结果集不一样.明显这个是一个bug.

修改如下:

select /*+ or_expand(@sel$1 (1) (2) (3) ) */ data
  from address
 where ( :choice = 1 and street = :val )
    or ( :choice = 2 and suburb = :val )
    or ( :choice = 3 and post_code = :val);

–//注使用提示/*+ or_expand(@sel$1 ) */结果是正确的.

scott@test01p> @ dpc ” outline    
plan hash value: 3525475520
—————————————————————————————————————————————————–
| id  | operation                              | name             | starts | e-rows |e-bytes| cost (%cpu)| e-time   | a-rows |   a-time   | buffers |
—————————————————————————————————————————————————–
|   0 | select statement                       |                  |      1 |        |       |   900 (100)|          |   1000 |00:00:00.02 |    1640 |
|   1 |  view                                  | vw_ore_b7380f92  |      1 |  11009 |  1096k|   900   (1)| 00:00:01 |   1000 |00:00:00.02 |    1640 |
|   2 |   union-all                            |                  |      1 |        |       |            |          |   1000 |00:00:00.02 |    1640 |
|*  3 |    filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |
|   4 |     table access by index rowid batched| address          |      0 |     10 |  1050 |    11   (0)| 00:00:01 |      0 |00:00:00.01 |       0 |
|*  5 |      index range scan                  | i_address_stress |      0 |     10 |       |     1   (0)| 00:00:01 |      0 |00:00:00.01 |       0 |
|*  6 |    filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |
|*  7 |     table access full                  | address          |      0 |  10000 |  1054k|   445   (1)| 00:00:01 |      0 |00:00:00.01 |       0 |
|*  8 |    filter                              |                  |      1 |        |       |            |          |   1000 |00:00:00.02 |    1640 |
|*  9 |     table access full                  | address          |      1 |    999 |   108k|   445   (1)| 00:00:01 |   1000 |00:00:00.02 |    1640 |
—————————————————————————————————————————————————–
query block name / object alias (identified by operation id):
————————————————————-
   1 – set$49e1c21b   / vw_ore_b7380f92@sel$b7380f92
   2 – set$49e1c21b
   3 – set$49e1c21b_1
   4 – set$49e1c21b_1 / address@sel$1
   5 – set$49e1c21b_1 / address@sel$1
   6 – set$49e1c21b_2
   7 – set$49e1c21b_2 / address@sel$1
   8 – set$49e1c21b_3
   9 – set$49e1c21b_3 / address@sel$1
outline data
————-
  /*+
      begin_outline_data
      ignore_optim_embedded_hints
      optimizer_features_enable(‘12.2.0.1’)
      db_version(‘12.2.0.1’)
      all_rows
      outline_leaf(@”set$49e1c21b_3″)
      outline_leaf(@”set$49e1c21b_2″)
      outline_leaf(@”set$49e1c21b_1″)
      outline_leaf(@”set$49e1c21b”)
      or_expand(@”sel$1″ (1) (2) (3))
      outline_leaf(@”sel$b7380f92″)
      outline(@”set$49e1c21b”)
      or_expand(@”sel$1″ (1) (2) (3))
      outline(@”sel$1″)
      no_access(@”sel$b7380f92″ “vw_ore_b7380f92″@”sel$b7380f92”)
      index_rs_asc(@”set$49e1c21b_1″ “address”@”sel$1” (“address”.”street”))
      batch_table_access_by_rowid(@”set$49e1c21b_1″ “address”@”sel$1”)
      full(@”set$49e1c21b_2″ “address”@”sel$1”)
      full(@”set$49e1c21b_3″ “address”@”sel$1”)
      end_outline_data
  */
peeked binds (identified by position):
————————————–
   2 – :2 (number): 6
   4 – :2 (number, primary=2)
   6 – :2 (number, primary=2)
predicate information (identified by operation id):
—————————————————
   3 – filter(:choice=1)
   5 – access(“street”=:val)
   6 – filter(:choice=2)
   7 – filter((“suburb”=:val and (lnnvl(:choice=1) or lnnvl(“street”=:val))))
   8 – filter(:choice=3)
   9 – filter((“post_code”=:val and (lnnvl(:choice=1) or lnnvl(“street”=:val)) and (lnnvl(:choice=2) or lnnvl(“suburb”=:val))))

3.测试使用use_concate看看:
set linesize 100
select /*+ use_concat */ data
  from address
 where ( :choice = 1 and street = :val )
    or ( :choice = 2 and suburb = :val )
    or ( :choice = 3 and post_code = :val);

scott@test01p> @ dpc ” outline

plan hash value: 2048882018
—————————————————————————————————————————————————-
| id  | operation                             | name             | starts | e-rows |e-bytes| cost (%cpu)| e-time   | a-rows |   a-time   | buffers |
—————————————————————————————————————————————————-
|   0 | select statement                      |                  |      1 |        |       |   900 (100)|          |   1000 |00:00:00.01 |    1640 |
|   1 |  concatenation                        |                  |      1 |        |       |            |          |   1000 |00:00:00.01 |    1640 |
|*  2 |   filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |
|   3 |    table access by index rowid batched| address          |      0 |     10 |  1110 |    11   (0)| 00:00:01 |      0 |00:00:00.01 |       0 |
|*  4 |     index range scan                  | i_address_stress |      0 |     10 |       |     1   (0)| 00:00:01 |      0 |00:00:00.01 |       0 |
|*  5 |   filter                              |                  |      1 |        |       |            |          |   1000 |00:00:00.01 |    1640 |
|*  6 |    table access full                  | address          |      1 |   1000 |   108k|   445   (1)| 00:00:01 |   1000 |00:00:00.01 |    1640 |
|*  7 |   filter                              |                  |      1 |        |       |            |          |      0 |00:00:00.01 |       0 |
|*  8 |    table access full                  | address          |      0 |   9999 |  1083k|   445   (1)| 00:00:01 |      0 |00:00:00.01 |       0 |
—————————————————————————————————————————————————-

query block name / object alias (identified by operation id):
————————————————————-

   1 – sel$1
   3 – sel$1_1 / address@sel$1
   4 – sel$1_1 / address@sel$1
   6 – sel$1_2 / address@sel$1_2
   8 – sel$1_3 / address@sel$1_3

outline data
————-

  /*+
      begin_outline_data
      ignore_optim_embedded_hints
      optimizer_features_enable(‘12.2.0.1’)
      db_version(‘12.2.0.1’)
      all_rows
      outline_leaf(@”sel$1″)
      outline_leaf(@”sel$1_1″)
      use_concat(@”sel$1″ 8 or_predicates(1) predicate_reorders((5 2) (6 3) (7 4) (8 5) (9 6) (10 7) (2 8) (4 9) (3 10)))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~      
–//始终不明白里面那一串表示什么?

      outline_leaf(@”sel$1_2″)
      outline_leaf(@”sel$1_3″)
      outline(@”sel$1″)
      index_rs_asc(@”sel$1_1″ “address”@”sel$1” (“address”.”street”))
      batch_table_access_by_rowid(@”sel$1_1″ “address”@”sel$1”)
      full(@”sel$1_2″ “address”@”sel$1_2”)
      full(@”sel$1_3″ “address”@”sel$1_3”)
      end_outline_data
  */

peeked binds (identified by position):
————————————–

   2 – :2 (number): 6
   4 – :2 (number, primary=2)
   6 – :2 (number, primary=2)

predicate information (identified by operation id):
—————————————————

   2 – filter(:choice=1)
   4 – access(“street”=:val)
   5 – filter(:choice=3)
   6 – filter((“post_code”=:val and (lnnvl(“street”=:val) or lnnvl(:choice=1))))
   7 – filter(:choice=2)
   8 – filter((“suburb”=:val and (lnnvl(:choice=3) or lnnvl(“post_code”=:val)) and (lnnvl(“street”=:val) or lnnvl(:choice=1))))

(0)
上一篇 2022年3月22日
下一篇 2022年3月22日

相关推荐