先從一個案例入手，開始深入理解學習Oracle的並行機制，並以此延申到其他數據庫。
1系統持續報direct path read，導致IO短時間內變化較大，經過定位為sql_id為4j4g5wsd7nbjs導致，由於該sql中並無對表
dayureport（199G）的過濾條件，當前並無優化空間。
select 'I' || ' | ' || to_char(sysdate, 'yyyymmdd') || ' | ' ||
car.car_caseid || ' | ' || car.car_txnseqno || ' | ' ||
car.rei_reporttype || ' | ' || car.car_userid || ' | ' ||
car.car_modifytime || ' | ' || car.car_modifyseq || ' | ' ||
car.rei_filename || ' | ' || car.rei_batchno || ' | ' || ca r.rei_extdate || ' | ' || car.rei_reportdate || ' | ' || car.rei_cati_seqno || ' | ' || car.rei_htcr_seqno || ' | ' || car.rei_tsdt_seqno || ' | ' || car.rei_ctif_seqno || ' | ' || car.rei_atif_seqno || ' | ' || car.rei_rpdi_seqno || ' | ' || car.rei_dtdt_seqno || ' | ' || car.rei_receipttype || ' | ' || car.rei_subtype || ' | ' || car.rei_cifbranchcd || ' | ' || car.rei_transloc || ' | ' || car.ety_rgn_code || ' | ' || car.car_customertype || ' | ' || car.txn_rgn_code || ' | ||'
from dayureport car
where car_caseid in
(select cad_caseid from cad_day_20180818 where ety_rgn_code = 42208);
2.在排查過程中發現表dayureport自帶degree屬性，
這
使sql即使沒有加並行參數，但實際執行過程中也起了50個並行度，其執行計划如下，其中:TQ為開啟並行標志。
Global Information
------------------------------
Status : DONE (ALL ROWS)
Instance ID : 1
Session : FINDNA (1751:64765)
SQL ID : 4j4g5wsd7nbjs
SQL Execution ID : 16777216
Execution Started : 08/23/2018 18:21:03
First Refresh Time : 08/23/2018 18:21:03
Last Refresh Time : 08/23/2018 18:22:43
Duration : 100s
Module/Action : SQL*Plus/-
Service : oraAMLM
Program : sqlplus@pamlaeadb1 (TNS V1-V3)
Fetch Calls : 250
Global Stats
=======================================================================================================
| Elapsed | Cpu | IO | Application | Concurrency | Other | Fetch | Buffer | Read | Read |
| Time(s) | Time(s) | Waits(s) | Waits(s) | Waits(s) | Waits(s) | Calls | Gets | Reqs | Bytes |
=======================================================================================================
| 4632 | 241 | 4263 | 0.00 | 0.00 | 128 | 250 | 26M | 2M | 194GB |
=======================================================================================================
Parallel Execution Details (DOP=50 , Servers Allocated=50) --degree of parallel
===========================================================================================================================================================
| Name | Type | Server# | Elapsed | Cpu | IO | Application | Concurrency | Other | Buffer | Read | Read | Wait Events |
| | | | Time(s) | Time(s) | Waits(s) | Waits(s) | Waits(s) | Waits(s) | Gets | Reqs | Bytes | (sample #) |
===========================================================================================================================================================
| PX Coordinator | QC | | 0.46 | 0.46 | | 0.00 | | | 986 | | . | |
| p000 | Set 1 | 1 | 54 | 3.25 | 49 | | | 1.67 | 353K | 21848 | 3GB | direct path read (48) |
| p001 | Set 1 | 2 | 94 | 4.86 | 87 | | 0.00 | 2.67 | 530K | 32831 | 4GB | direct path read (89) |
| p002 | Set 1 | 3 | 95 | 4.93 | 87 | | | 2.54 | 535K | 33108 | 4GB | direct path read (88) |
| p003 | Set 1 | 4 | 94 | 4.92 | 86 | | | 2.70 | 531K | 32890 | 4GB | direct path read (88) |
| p004 | Set 1 | 5 | 56 | 3.44 | 51 | | | 1.88 | 376K | 23314 | 3GB | direct path read (47) |
| p005 | Set 1 | 6 | 94 | 4.83 | 87 | | | 2.48 | 526K | 32485 | 4GB | direct path read (83) |
| p006 | Set 1 | 7 | 94 | 4.86 | 86 | | | 2.52 | 525K | 32492 | 4GB | direct path read (85) |
| p007 | Set 1 | 8 | 94 | 4.88 | 87 | | | 2.67 | 527K | 32718 | 4GB | direct path read (88) |
| p008 | Set 1 | 9 | 94 | 4.81 | 86 | | | 2.62 | 525K | 32432 | 4GB | direct path read (86) |
| p009 | Set 1 | 10 | 94 | 4.91 | 86 | | | 2.68 | 528K | 32772 | 4GB | direct path read (84) |
| p00a | Set 1 | 11 | 94 | 4.90 | 86 | | | 2.68 | 529K | 32716 | 4GB | direct path read (84) |
| p00b | Set 1 | 12 | 94 | 4.81 | 87 | | | 2.47 | 530K | 32743 | 4GB | direct path read (86) |
| p00c | Set 1 | 13 | 94 | 4.84 | 87 | | | 2.48 | 528K | 32769 | 4GB | direct path read (86) |
| p00d | Set 1 | 14 | 95 | 4.85 | 87 | | | 2.64 | 533K | 33007 | 4GB | direct path read (85) |
| p00e | Set 1 | 15 | 94 | 4.81 | 86 | | | 2.65 | 529K | 32649 | 4GB | direct path read (87) |
| p00f | Set 1 | 16 | 94 | 4.85 | 87 | | | 2.50 | 531K | 32909 | 4GB | direct path read (86) |
| p00g | Set 1 | 17 | 94 | 4.91 | 87 | | | 2.66 | 530K | 32889 | 4GB | direct path read (88) |
| p00h | Set 1 | 18 | 94 | 4.87 | 87 | | | 2.50 | 534K | 32903 | 4GB | direct path read (80) |
| p00i | Set 1 | 19 | 94 | 5.01 | 86 | | | 2.58 | 529K | 32691 | 4GB | direct path read (84) |
| p00j | Set 1 | 20 | 94 | 4.83 | 86 | | | 2.64 | 528K | 32720 | 4GB | direct path read (86) |
| p00k | Set 1 | 21 | 94 | 4.93 | 86 | | | 2.69 | 533K | 32979 | 4GB | direct path read (86) |
| p00l | Set 1 | 22 | 94 | 4.82 | 87 | | | 2.49 | 526K | 32622 | 4GB | direct path read (87) |
| p00m | Set 1 | 23 | 96 | 4.94 | 88 | | | 2.69 | 536K | 33214 | 4GB | direct path read (87) |
| p00n | Set 1 | 24 | 94 | 4.86 | 86 | | | 2.68 | 533K | 32870 | 4GB | direct path read (87) |
| p00o | Set 1 | 25 | 94 | 5.01 | 86 | | | 2.78 | 527K | 32605 | 4GB | direct path read (85) |
| p00p | Set 1 | 26 | 95 | 4.89 | 88 | | | 2.68 | 536K | 33159 | 4GB | direct path read (89) |
| p00q | Set 1 | 27 | 94 | 4.83 | 87 | | | 2.63 | 534K | 32984 | 4GB | direct path read (89) |
| p00r | Set 1 | 28 | 95 | 4.89 | 88 | | | 2.52 | 535K | 33126 | 4GB | direct path read (87) |
| p00s | Set 1 | 29 | 94 | 4.85 | 87 | | | 2.64 | 525K | 32614 | 4GB | direct path read (88) |
| p00t | Set 1 | 30 | 94 | 4.85 | 87 | | | 2.50 | 529K | 32742 | 4GB | direct path read (84) |
| p00u | Set 1 | 31 | 94 | 4.86 | 86 | | | 2.67 | 527K | 32587 | 4GB | direct path read (88) |
| p00v | Set 1 | 32 | 94 | 4.83 | 87 | | | 2.46 | 530K | 32747 | 4GB | direct path read (83) |
| p00w | Set 1 | 33 | 94 | 4.75 | 87 | | | 2.61 | 524K | 32413 | 4GB | direct path read (83) |
| p00x | Set 1 | 34 | 94 | 4.91 | 87 | | | 2.69 | 532K | 32937 | 4GB | direct path read (86) |
| p00y | Set 1 | 35 | 94 | 5.01 | 87 | | | 2.57 | 538K | 33278 | 4GB | direct path read (86) |
| p00z | Set 1 | 36 | 94 | 4.91 | 87 | | | 2.52 | 533K | 32933 | 4GB | direct path read (84) |
| p010 | Set 1 | 37 | 94 | 4.77 | 87 | | | 2.62 | 523K | 32298 | 4GB | direct path read (90) |
| p011 | Set 1 | 38 | 94 | 4.85 | 87 | | | 2.50 | 529K | 32747 | 4GB | direct path read (81) |
| p012 | Set 1 | 39 | 95 | 4.94 | 87 | | | 2.71 | 531K | 32932 | 4GB | direct path read (87) |
| p013 | Set 1 | 40 | 94 | 4.88 | 86 | | | 2.51 | 528K | 32748 | 4GB | direct path read (89) |
| p014 | Set 1 | 41 | 94 | 4.82 | 86 | | | 2.47 | 529K | 32762 | 4GB | direct path read (87) |
| p015 | Set 1 | 42 | 95 | 4.95 | 87 | | | 2.74 | 536K | 33190 | 4GB | direct path read (86) |
| p016 | Set 1 | 43 | 94 | 4.86 | 87 | | | 2.50 | 533K | 33021 | 4GB | direct path read (89) |
| p017 | Set 1 | 44 | 94 | 4.88 | 87 | | | 2.67 | 532K | 32987 | 4GB | direct path read (82) |
| p018 | Set 1 | 45 | 94 | 4.86 | 86 | | | 2.50 | 527K | 32681 | 4GB | direct path read (87) |
| p019 | Set 1 | 46 | 94 | 4.88 | 86 | | | 2.53 | 529K | 32783 | 4GB | direct path read (85) |
| p01a | Set 1 | 47 | 95 | 4.97 | 87 | | | 2.56 | 530K | 32894 | 4GB | direct path read (87) |
| p01b | Set 1 | 48 | 94 | 4.94 | 86 | | | 2.70 | 528K | 32673 | 4GB | direct path read (86) |
| p01c | Set 1 | 49 | 95 | 4.92 | 87 | | | 2.52 | 524K | 32512 | 4GB | direct path read (88) |
| p01d | Set 1 | 50 | 95 | 4.83 | 87 | | | 2.62 | 531K | 32981 | 4GB | direct path read (86) |
===========================================================================================================================================================
SQL Plan Monitoring Details (Plan Hash Value=3281332499)
===============================================================================================================================================================================
| Id | Operation | Name | Rows | Cost | Time | Start | Execs | Rows | Read | Read | Mem | Activity | Activity Detail |
| | | | (Estim) | | Active(s) | Active | | (Actual) | Reqs | Bytes | (Max) | (%) | (# samples) |
===============================================================================================================================================================================
| 0 | SELECT STATEMENT | | | | 5 | +96 | 51 | 3721 | | | . | | |
| 1 | PX COORDINATOR | | | | 5 | +96 | 51 | 3721 | | | . | | |
| 2 | PX SEND QC (RANDOM) | :TQ10001 | 208K | 111K | 3 | +96 | 50 | 3721 | | | . | | |
| 3 | HASH JOIN | | 208K | 111K | 100 | +1 | 50 | 3721 | | | 59MB | 0.11 | Cpu (5) |
| 4 | BUFFER SORT | | | | 1 | +2 | 50 | 72800 | | | 3MB | | |
| 5 | JOIN FILTER CREATE | :BF0000 | 1551 | 10 | 1 | +2 | 50 | 72800 | | | . | | |
| 6 | PX RECEIVE | | 1551 | 10 | 1 | +2 | 50 | 72800 | | | . | | |
| 7 | PX SEND BROADCAST | :TQ10000 | 1551 | 10 | 1 | +2 | 1 | 72800 | | | . | | |
| 8 | PARTITION LIST SINGLE | | 1551 | 10 | 1 | +2 | 1 | 1456 | | | . | | |
| 9 | TABLE ACCESS FULL | CAD_DAY_20180818 | 1551 | 10 | 1 | +2 | 1 | 1456 | | | . | | |
| 10 | JOIN FILTER USE | :BF0000 | 1G | 111K | 99 | +2 | 50 | 29M | | | . | | |
| 11 | PX BLOCK ITERATOR | | 1G | 111K | 99 | +2 | 50 | 29M | | | . | | |
| 12 | TABLE ACCESS FULL | CAR_CASEREPORT | 1G | 111K | 100 | +1 | 20606 | 29M | 2M | 194GB | . | 99.89 | Cpu (375) |
| | | | | | | | | | | | | | direct path read (4227) |
===============================================================================================================================================================================
3.表dayureport帶有並行degree的原因是由於我們在建表的時候指定了並行參數，導致表的degree發生變化。如以下測試：
create table parallel_test1 parallel (degree 4) as select * from dba_objects;
SQL> select DEGREE from dba_tables where table_name='PARALLEL_TEST1';
TABLE_NAME DEGREE
---------------------------------------- ------------------------------
PARALLEL_TEST1 4 --可見degree受創建表時指定的degree影響。
create table parallel_test2 parallel (degree 8) as select * from dba_objects;
SQL> select table_name,degree from dba_tables where table_name='PARALLEL_TEST2';
TABLE_NAME DEGREE
---------------------------------------- ------------------------------
PARALLEL_TEST2 8
4.表自帶並行度在對表的並發訪問個數較多時，會造成系統瞬間並發激增，系統負載瞬時壓力過大的情況。可通過以下方式調整並行度，見如下測試：
SQL> alter table PARALLEL_TEST2 parallel;
Table altered.
SQL> select DEGREE from dba_tables where table_name='PARALLEL_TEST2';
DEGREE
------------------------------
DEFAULT
5.調整默認並行度為是不是DDL操作呢？
SQL> select LAST_DDL_TIME from dba_objects where object_name='PARALLEL_TEST2';
LAST_DDL_TIME
--------------
20180823 21:35
SQL> alter table PARALLEL_TEST2 parallel 1; --此處建議將其默認並行度改為1，而不是默認值
Table altered.
SQL> select DEGREE from dba_tables where table_name='PARALLEL_TEST2';
DEGREE
------------------------------
DEFAULT --
SQL> select LAST_DDL_TIME from dba_objects where object_name='PARALLEL_TEST2';
LAST_DDL_TIME
--------------
20180824 09:37
5.當degree的並行度為默認時，其default值為多少？
當采用默認值時，default=cpu_count*parallel_thread_per_cpu。也就是說當表的默認度為default時，其並行度為4所以設置這個參數的時候要謹慎！
SQL> show parameter cpu_count
NAME TYPE VALUE
------------------------------------ --------------------------------- ------------------------------
cpu_count integer 2
SQL> show parameter parallel_thread
NAME TYPE VALUE
------------------------------------ --------------------------------- ------------------------------
parallel_threads_per_cpu integer 2