Why I have an execution plan
Hello
I do not understand why this sql has a bad execution plan?
SELECT DISTINCT LETTER0_. LE_IDENT AS col_0_0_
THE LETTER LETTER0_
LEFT OUTER JOIN CONTEXT1_ CONTEXT
ON LETTER0_. LE_PTRCTXID = CONTEXT1_. CTX_IDENT
LEFT OUTER JOIN F_DOC f_doc2_
ON LETTER0_. LE_PTRDOCID = f_doc2_. DOC_IDENT
WHERE LETTER0_. LE_IDENT IN
(SELECT LETTER3_. LE_IDENT
THE LETTER LETTER3_, CONTEXT CONTEXT4_
WHERE LETTER3_. LE_PTRCTXID = CONTEXT4_. CTX_IDENT
AND CONTEXT4_. CTX_PTRPOLID = 400728434)
OR LETTER0_. LE_IDENT IN (2525432);
Execution plan
----------------------------------------------------------
0 SELECT optimizer Mode STATEMENT = ALL_ROWS (cost = card 175 = 14 K bytes = 83 K)
1 FILTER 0
2 1 INDEX FULL SCAN PROSHIVA1. PK_LE_CLE (cost = card 175 = 284 K bytes = 1 M)
3 1 LOOPS IMBRIQUEES (cost = 4 cards = 1 bytes = 22)
TABLE 3 ACCESS BY INDEX ROWID PROSHIVA1 4. F_LETTRES (cost = 2 card = 1 bytes = 13)
5 4 INDEX UNIQUE PROSHIVA1 SCAN. PK_LE_CLE (cost = 1 card = 1)
6 TABLE ACCESS BY INDEX ROWID PROSHIVA1 3. F_CONTEXT (cost = 2 card = 1 bytes = 9)
7 6 INDEX RANGE SCAN PROSHIVA1. IND_CTX_PTRPOLID (cost = 1 card = 1)
Statistics
----------------------------------------------------------
5 user calls
0 physical total bytes read
physical write 0 total bytes
0 3 spare statistics
0 commit failures of drain plug: can not pin
Extension of TBS 0: extended bytes
0 total number of times SMON post
SMON 0 posted for undo segment recovery
SMON 0 posted to drop the temp segment
prealloc segment 0 tasks
1 rows processed
9f9cf8b7-e5bc-4AAF-B369-b189dadb9d77 wrote:
Thanks for the reply
due to the second line of the execution plan. INDEX FULL SCAN PROSHIVA1. PK_LE_CLE (cost = card 175 = 284 K bytes = 1 M).
This sql takes 2 seconds. and if I remove any OF the sql condition. There are only 200 ms.
This is a typical problem optimizer - IN / associate EXISTS OR does work well in many cases. Most likely, if you simply specify the IN clause, the optimizer will unnest subquery, which could be much faster in this case.
Furthermore, the optimizer has already done a good job in eliminating unnecessary outer joins, there is no mention of CONTEXT or F_DOC in the source line of the FILTER operator. He even eliminated the SEPARATE, most likely because he knows that LETTER. LE_IDENT is unique.
Think of what the database has to do with your request - need to find matches which have the mentioned ID or who have a match in the specified list as subquery. So without more clever query processing, it must pass the primary key index set and check, for each line, if it is a match in the IN subquery - that's exactly what makes the FILTER operator, it executes the LOOP IMBRIQUEE join that represents the IN clause for each row returned by the full index scan.
Ideally, this should be transformed by using a transformation of CONCATENATION, where this query is actually split in two: part of the research for the mentioned ID and the other party responsible for looking for matches with subqueries and this making, ensuring that the two branches do not result in duplicate (since a line could satisfy both conditions).
You can try an explicit indication of USE_CONCAT for the query, but I doubt that the optimizer will do it automatically.
So if it does not, you could do the same thing as a manual rewrite:
SELECT LETTER0_. LE_IDENT AS col_0_0_
THE LETTER LETTER0_
WHERE LETTER0_. LE_IDENT = 2525432
UNION ALL
SELECT LETTER0_. LE_IDENT AS col_0_0_
THE LETTER LETTER0_
WHERE LETTER0_. LE_IDENT IN
(SELECT LETTER3_. LE_IDENT
THE LETTER LETTER3_, CONTEXT CONTEXT4_
WHERE LETTER3_. LE_PTRCTXID = CONTEXT4_. CTX_IDENT
AND CONTEXT4_. CTX_PTRPOLID = 400728434)
AND LNNVL (LETTER0_. LE_IDENT = 2525432);
The LNNVL is the bit that filter potential duplicates and which would be automatically generated by the transformation of the CONCATENATION.
Randolf
Tags: Database
Similar Questions
-
Why sql_trace sometimes generate an execution plan
I use "alter session set sql_trace = true;" to trace a package.
I use the tkprof to format the raw trace with no options.
But there's one thing confuses me.
In the trace of any statement have plans of execution as below.
SELECT BATCH_ID FROM OPS_DW_BUSINESS_DT WHERE CURRENT_I = 'Y' AND OPS_NAME = :B1 call count cpu elapsed disk query current rows ------- ------ -------- ---------- ---------- ---------- ---------- ---------- Parse 1 0.00 0.00 0 0 0 0 Execute 2 0.00 0.00 0 0 0 0 Fetch 2 0.00 0.00 1 2 0 2 ------- ------ -------- ---------- ---------- ---------- ---------- ---------- total 5 0.00 0.00 1 2 0 2 Misses in library cache during parse: 1 Misses in library cache during execute: 1 Optimizer mode: ALL_ROWS Parsing user id: 61 (recursive depth: 1) Rows Row Source Operation ------- --------------------------------------------------- 2 INDEX RANGE SCAN OPS_DW_BUSINESS_DT_IDX1 (cr=2 pr=1 pw=0 time=350 us)(object id 135871)
Any statement have no implementation plan
******************************************************************************** SELECT ACTIVITY_CODE,TRIP,EQUIPMENT_N,LOCATION,DISC_VSL_NAME,DISC_VOY, LOAD_VSL_NAME,LOAD_VOY,MOVE_OPS, ACTIVITY_DT, ACT_SEQ, LAG(TRIP, 1) OVER(ORDER BY TRIP, ACTIVITY_DT, ACT_SEQ) AS PREV_TRIP, LEAD(TRIP, 1) OVER(ORDER BY TRIP, ACTIVITY_DT, ACT_SEQ) AS NEXT_TRIP FROM OIS_HRLY_PMOPS B WHERE ( BATCH_ID = :B3 AND (INSTR(:B2 , ACTIVITY_CODE) <> 0 OR INSTR(:B1 , ACTIVITY_CODE) <> 0) ) OR ( BATCH_ID < :B3 AND (INSTR(:B2 , ACTIVITY_CODE) <> 0 OR INSTR(:B1 , ACTIVITY_CODE) <> 0) AND EXISTS (SELECT 1 FROM OIS_HRLY_PMOPS C WHERE C.BATCH_ID = :B3 AND C.TRIP=B.TRIP) ) ORDER BY TRIP, ACTIVITY_DT, ACT_SEQ call count cpu elapsed disk query current rows ------- ------ -------- ---------- ---------- ---------- ---------- ---------- Parse 1 0.00 0.00 0 0 0 0 Execute 1 0.02 0.01 0 0 0 0 Fetch 2246 25.77 24.87 0 46319 0 224517 ------- ------ -------- ---------- ---------- ---------- ---------- ---------- total 2248 25.79 24.88 0 46319 0 224517 Misses in library cache during parse: 1 Misses in library cache during execute: 1 Optimizer mode: ALL_ROWS Parsing user id: 61 (recursive depth: 1) ********************************************************************************
I know that I can use the tkprof explain = name of user and password to force generate a. But for the time being. Why some statement make execution plan while others have not?
Concerning
Monique
> Source line operation lines
What you have is Source line operations - the * real * execution plan.
The EXPLAIN command option gives you the estimated execution plan.
Why didn't you not operations Source online for all SQLs? Row Source operations are printed only if the cursor is closed. Usually in pure SQL sessions, SQL cursor is closed when the following SQL is running. However, with PLSQL and stored procedures, the procedure cannot close the cursor for the SQL statement, it calls (closing is made when the session is closed). In such a case, the operations line Source not to print.
Hemant K Collette
-
Why I have two different execution plans for the same query on two different servers
Hello everyone.
I need your help to solve the problem quickly.
In a nutshell, we have two servers that have the same version of Oracle RDBMS (11.2.0.4 EE). One of them for purposes of development and another is a production one.
We have therefore two different execution plans for the same query executed on both servers. The only case of execution is OK and another is too slow.
So I have to slow down the work of query using the same scheme to explain that young.
Fence wire.
-
Hello world
I have a little a problem of performance on 12 c that gives me a little trouble at the head. I moved from 11 to 12 databases and no amendment of the application have been made. Our requests are generated somewhat dynamically, so that they are the same thing every time.
Let's start with the execution plan I get:
SQL > select * from table (dbms_xplan.display ());
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Hash value of plan: 3567104424
-----------------------------------------------------------------------------------------------------------------------------------------------------------
| ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time | TQ | IN-OUT | PQ Distrib.
-----------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 55. 7095 | 3764 (1) | 00:00:01 | | | |
| 1. COORDINATOR OF PX | | | | | | | | |
| 2. PX SEND QC (ORDER). : TQ10006 | 55. 7095 | 3764 (1) | 00:00:01 | Q1, 06 | P > S | QC (ORDER).
| 3. SORT ORDER BY | | 55. 7095 | 3764 (1) | 00:00:01 | Q1, 06 | SVCP | |
| 4. PX RECEIVE | | 55. 7095 | 3763 (1) | 00:00:01 | Q1, 06 | SVCP | |
| 5. RANGE OF SEND PX | : TQ10005 | 55. 7095 | 3763 (1) | 00:00:01 | Q1, 05 | P > P | RANGE |
| 6. UNIQUE FATE | | 55. 7095 | 3763 (1) | 00:00:01 | Q1, 05 | SVCP | |
|* 7 | HASH JOIN | | 55. 7095 | 3762 (1) | 00:00:01 | Q1, 05 | SVCP | |
| 8. PX RECEIVE | | 801 | 50463 | 3696 (1) | 00:00:01 | Q1, 05 | SVCP | |
| 9. PX SEND HASH | : TQ10003 | 801 | 50463 | 3696 (1) | 00:00:01 | Q1, 03 | P > P | HASH |
| * 10 | HASH JOIN | | 801 | 50463 | 3696 (1) | 00:00:01 | Q1, 03 | SVCP | |
| 11. RECEIVE PX | | 801 | 40851 | 2333 (1) | 00:00:01 | Q1, 03 | SVCP | |
| 12. PX SEND BROADCAST | : TQ10002 | 801 | 40851 | 2333 (1) | 00:00:01 | Q1, 02 | P > P | BROADCAST |
| 13. NESTED LOOPS | | 801 | 40851 | 2333 (1) | 00:00:01 | Q1, 02 | SVCP | |
| 14. KIND OF BUFFER. | | | | | Q1, 02 | ISSUE | |
| 15. RECEIVE PX | | | | | | Q1, 02 | SVCP | |
| 16. PX SEND HASH | : TQ10000 | | | | | | S > P | HASH |
| 17. NESTED LOOPS | | 823. 31274 | 1509 (1) | 00:00:01 | | | |
| * 18. TABLE ACCESS BY ROWID INDEX BATCH | PAGED_LOOKUP_PKS | 500 | 9500 | 3 (0) | 00:00:01 | | | |
| * 19. INDEX RANGE SCAN | PAGED_LOOKUP_PKS_IDX2 | 1. | 2 (0) | 00:00:01 | | | |
| 20. TABLE ACCESS BY ROWID INDEX BATCH | BILL_ITEM | 2. 38. 4 (0) | 00:00:01 | | | |
| * 21. INDEX RANGE SCAN | BILL_ITEM_FK2 | 4. | 2 (0) | 00:00:01 | | | |
| * 22. INDEX UNIQUE SCAN | PK_INSERTION | 1. 13. 1 (0) | 00:00:01 | Q1, 02 | SVCP | |
| 23. ITERATOR BLOCK PX | | 1548K | 17 M | 1353 (2) | 00:00:01 | Q1, 03 | ISSUE | |
| 24. FULL RESTRICTED INDEX SCAN FAST | BOOKING_ACCOUNT_1 | 1548K | 17 M | 1353 (2) | 00:00:01 | Q1, 03 | SVCP | |
| 25. PX RECEIVE | | 22037 | 1420K | 65 (2) | 00:00:01 | Q1, 05 | SVCP | |
| 26. PX SEND HASH | : TQ10004 | 22037 | 1420K | 65 (2) | 00:00:01 | Q1, 04 | S > P | HASH |
| 27. SELECTOR PX | | | | | | Q1, 04 | SCWC | |
| 28. TABLE ACCESS FULL | CONTACT | 22037 | 1420K | 65 (2) | 00:00:01 | Q1, 04 | SCWP | |
-----------------------------------------------------------------------------------------------------------------------------------------------------------
Information of predicates (identified by the operation identity card):
---------------------------------------------------
7 - access ("ACCOUNT_ID" ="ACCOUNT_ID")
10 - access ("BOOKING" ="BOOKING")
18 - filter("T1".") SEQUENCE_NO' < 501 AND "T1". ("' SEQUENCE_NO" > = 1).
19 - access("T1".") SESSION_ID '= 123456 AND 'T1'.' SEARCH_ID "= 25)
21 - access("T1".") N1 "=" BILL_ID")
22 - access ("BOOKING" = "BOOKING" AND "INSERTION_SET" = "INSERTION_SET" AND "INSERT"="INSERT")
Note
-----
-the dynamic statistics used: dynamic sampling (level = 2)
-This is an adaptation plan
-2 directives Plan Sql used for this statement
51 selected lines.
Elapsed time: 00:00:00.15
SQL > spool off
OK, now let's go through the problem:
- It's a development running on a virtual server, and which hosts a few other databases, so the parallel execution is not a good thing. parallel_degree_policy is set to MANUAL, parallel_max_servers and all other parallel_ limits are set to 1 and tables have been changed with the settings of NOPARALLEL. So why is the execution plan always generated with all stages of parallel execution? I don't seem to get rid of in 12 c
- Next mystery is that the said plan of the explain command is an adaptation plan, and yet I put the true optimizer_adaptive_reproting_only
- Now to the problem of effective enforcement, so I'm playing around with all these settings. The query runs for 3-4 seconds, returning around about 500 cases. However, in some cases this same query with the same input variable races for hours and if I can believe the AWR and ASH reports, read a good 180 GB of data. The main wait event is direct path read temp temp and writing.
This is not isolated to that one query. I have a few queries now that all display the same behavior, one of them running overnight. I don't seem to get to a standard nested loop execution plans.
The entire base is a database plug-in and I don't know I just missed something in the new features Guide.
Would appreciate some ideas.
Thank you
If you want to disable parallel execution, you must set parallel_max_servers to zero. Maybe the optimizer thinks he can use a parallel plan because parallel_max_servers is non-zero (even though the number of slaves available means that it will be serialized to a parallel plan).
Note that you have a ticket saying dynamic stats have been used. Maybe you have a 11 for optimizer_dynamic_sampling setting, and allowing Oracle to be very inventive with collection of samples and parallelism.
You have also 2 SQL instructions in game. These are the things that get associated with objects rather than the instructions, then perhaps someone has been playing with parallelism and managed to associate the parallelism with one of the tables in your query (I am not sure 100% that it is possible, just throw a suggestion). Take a look at the SQL used for education guidelines.
To give us a little more information, you can:
Shoot memory execution plan dbms_xplan.display_cursor ({sql_id}, {number of children}, 'ALL'));
We show all the parallel settings (see setting the parallel)
Pull on the parameters of the optimizer for query memory (select name, value of V$ sql_optimizer_env where sql_id = {your sql identifier} and child_number = {your child number})
Concerning
Jonathan Lewis
-
Should I wait until the end of the execution time of the query for the execution plan?
Hello Experts,
I want to see the execution plan of the query below. However, it takes more than 3 hours. Should I wait all the time to see the execution plan?
Note: EXPLAIN PLAN for does not work. (I mean that I do not see the actual line number, etc. with EXPLAIN the PLAN of market)
You can see the output of the execution plan when I canceled the execution after 1 minute.
My first question is: what should I do to see the execution plan for queries running out of time time?
2nd question: when I cancel the query during execution in order to see the execution plan, will I see the specific plan of execution or erroneous values? Because the first execution plan seems inaccurate, what do you think?
question 3: why EXPLAIN the PLAN for the clause does not work? Also, should I use EXPLAIN the PLAN of the clause to this scenerio? Can I see the result of running for long time without her queries?
Thnaks for your help.
Oracle Database 11 g Enterprise Edition Release 11.2.0.2.0 - 64 bit Production
PL/SQL Release 11.2.0.2.0 - Production
CORE Production 11.2.0.2.0
AMT for Linux: Version 11.2.0.2.0 - Production
NLSRTL Version 11.2.0.2.0 - Production
Select / * + GATHER_PLAN_STATISTICS NO_PARALLEL * / J.INVOICEACCOUNT, J.INVOICEID, J.INVOICEDATE, (T.LINEAMOUNT + T.LINEAMOUNTTAX) price
of custinvoicejour j join custinvoicetrans t on
substr (nls_lower (j.DataAreaId), 1, 7) = substr (nls_lower (t.dataareaid), 1, 7) and
substr (nls_lower (J.INVOICEID), 1: 25) = substr (nls_lower (t.INVOICEID), 1: 25)
where
substr (nls_lower (T.DATAAREAID), 1, 7) = '201' and T.AVBROCHURELINENUM = 29457
and substr (nls_lower (j.dataareaid), 1, 7) = '201' and
J.INVOICEACCOUNT in
(select IT. Drmpos.avtr_seg_cust_campend ACCOUNTNUM this where THIS. CAMPAIGN = '201406' and THIS. SEGMENT_LEVEL in (', 'E'))
and J.AVAWARDSALES > 190
and substr (nls_lower (J.AVBILLINGCAMPAIGN), 1, 13) = '201406'
"and J.INVOICEDATE between ' 04.06.2014' and ' 13.06.2014 ';
SQL_ID, dznya6x7st0t8, number of children 0
-------------------------------------
Select / * + GATHER_PLAN_STATISTICS NO_PARALLEL * / J.INVOICEACCOUNT,.
J.INVOICEID, J.INVOICEDATE, (T.LINEAMOUNT + T.LINEAMOUNTTAX) price of
CustInvoiceJour j join custinvoicetrans t on
substr (nls_lower (j.DataAreaId), 1, 7) =
substr (nls_lower (t.DataAreaId), 1, 7) and
= substr (nls_lower (J.INVOICEID), 1: 25)
substr (nls_lower (t.INVOICEID), 1: 25) where
substr (nls_lower (T.DATAAREAID), 1, 7) = '201' and T.AVBROCHURELINENUM =
29457 and substr (nls_lower, (j.dataareaid), 1, 7) = '201' and
J.INVOICEACCOUNT in (select CE. ACCOUNTNUM of
drmpos.avtr_seg_cust_campend this where THIS. CAMPAIGN = '201406' and
IT. SEGMENT_LEVEL in (', 'E')) and J.AVAWARDSALES > 190 and
substr (nls_lower (J.AVBILLINGCAMPAIGN), 1, 13) = '201406' and
"J.INVOICEDATE between ' 04.06.2014' and ' 13.06.2014 '.
Hash value of plan: 2002317666
--------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID | Operation | Name | Begins | E - lines. A - lines. A - time | Pads | Bed | OMem | 1Mem | Used Mem.
--------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1. | 0 | 00:00:00.01 | 0 | 0 | | | |
|* 1 | HASH JOIN | | 1. 3956. 0 | 00:00:00.01 | 0 | 0 | 2254K | 1061K | 2190K (0) |
|* 2 | HASH JOIN | | 1. 87. 16676. 00:00:01.64 | 227K | 3552. 3109K | 1106K | 4111K (0) |
|* 3 | TABLE ACCESS BY INDEX ROWID | CUSTINVOICEJOUR | 1. 1155 | 31889 | 00:00:01.16 | 223KO | 15. | | |
|* 4 | INDEX RANGE SCAN | I_062INVOICEDATEORDERTYPEIDX | 1. 4943 | 134K | 00:00:00.83 | 45440 | 0 | | | |
| 5. SIMPLE LIST OF PARTITION. | 1. 82360 | 173K | 00:00:00.08 | 3809 | 3537 | | | |
|* 6 | TABLE ACCESS FULL | AVTR_SEG_CUST_CAMPEND | 1. 82360 | 173K | 00:00:00.06 | 3809 | 3537 | | | |
| 7. TABLE ACCESS BY INDEX ROWID | CUSTINVOICETRANS | 1. 4560 | 0 | 00:00:00.01 | 0 | 0 | | | |
|* 8 | INDEX RANGE SCAN | I_064INVLINENUMCAMPAIGNOFPRICE | 1. 4560 | 0 | 00:00:00.01 | 0 | 0 | | | |
--------------------------------------------------------------------------------------------------------------------------------------------------------------
Information of predicates (identified by the operation identity card):
---------------------------------------------------
1 - access("J".") "SYS_NC00299$"="T". "' SYS_NC00165$ ' AND SUBSTR (NLS_LOWER ('J'. "" "" REFFACTURE")(, 1, 25) = SUBSTR (NLS_LOWER ("T"." "" "REFFACTURE")(, 1, 25)).
2 - access("J".") INVOICEACCOUNT '= SYS_OP_C2C ("EC". ". ACCOUNTNUM'))
3 - filter("J".") AVAWARDSALES"> 190)
4 - access("J".") SYS_NC00299$ "= U ' 201"AND "J". INVOICEDATE"> = TO_DATE(' 2014-06-04 00:00:00', 'syyyy-mm-dd hh24:mi:ss') AND
"J"." SYS_NC00307$ "= U ' 201406"AND "J". INVOICEDATE"< = TO_DATE (' 2014-06-13 00:00:00 ',' syyyy-mm-dd hh24:mi:ss')))
filter ((' J'. "INVOICEDATE' > = 'J' AND TO_DATE(' 2014-06-04 00:00:00', 'syyyy-mm-dd hh24:mi:ss') '." " SYS_NC00307$ "= U '201406' AND"
"J"." INVOICEDATE"< = TO_DATE (' 2014-06-13 00:00:00 ',' syyyy-mm-dd hh24:mi:ss'))))
6 filter (("CE". "SEGMENT_LEVEL" = A "OR"THIS"." SEGMENT_LEVEL "=" E"))
8 - access("T".") SYS_NC00165$ "= U ' 201"AND "T". AVBROCHURELINENUM "= 29457)
filter ("T". ("AVBROCHURELINENUM" = 29457)
EXPLAIN PLAN FOR
Select / * + GATHER_PLAN_STATISTICS NO_PARALLEL * / J.INVOICEACCOUNT, J.INVOICEID, J.INVOICEDATE, (T.LINEAMOUNT + T.LINEAMOUNTTAX) price
of custinvoicejour j join custinvoicetrans t on
substr (nls_lower (j.DataAreaId), 1, 7) = substr (nls_lower (t.dataareaid), 1, 7) and
substr (nls_lower (J.INVOICEID), 1: 25) = substr (nls_lower (t.INVOICEID), 1: 25)
where
substr (nls_lower (T.DATAAREAID), 1, 7) = '201' and T.AVBROCHURELINENUM = 29457
and substr (nls_lower (j.dataareaid), 1, 7) = '201' and
J.INVOICEACCOUNT in
(select IT. Drmpos.avtr_seg_cust_campend ACCOUNTNUM this where THIS. CAMPAIGN = '201406' and THIS. SEGMENT_LEVEL in (', 'E'))
and J.AVAWARDSALES > 190
and substr (nls_lower (J.AVBILLINGCAMPAIGN), 1, 13) = '201406'
"and J.INVOICEDATE between ' 04.06.2014' and ' 13.06.2014 ';
SELECT * FROM table (DBMS_XPLAN. DISPLAY_CURSOR);
SELECT * FROM table (DBMS_XPLAN. DISPLAY_CURSOR ('7h1nbzqjgwsp7', 2));
SQL_ID, 7h1nbzqjgwsp7, number of children 2
EXPLAIN PLAN for select / * + GATHER_PLAN_STATISTICS NO_PARALLEL * /.
J.INVOICEACCOUNT, J.INVOICEID, J.INVOICEDATE,
(T.LINEAMOUNT + T.LINEAMOUNTTAX) join price j custinvoicejour
CustInvoiceTrans t on substr (nls_lower (j.dataareaid), 1, 7) =
substr (nls_lower (t.DataAreaId), 1, 7) and
= substr (nls_lower (J.INVOICEID), 1: 25)
substr (nls_lower (t.INVOICEID), 1: 25) where
substr (nls_lower (T.DATAAREAID), 1, 7) = '201' and T.AVBROCHURELINENUM =
29457 and substr (nls_lower, (j.dataareaid), 1, 7) = '201' and
J.INVOICEACCOUNT in (select CE. ACCOUNTNUM of
drmpos.avtr_seg_cust_campend this where THIS. CAMPAIGN = '201406' and
IT. SEGMENT_LEVEL in (', 'E')) and J.AVAWARDSALES > 190 and
substr (nls_lower (J.AVBILLINGCAMPAIGN), 1, 13) = '201406' and
"J.INVOICEDATE between ' 04.06.2014' and ' 13.06.2014 '.
NOTE: cannot fetch SQL_ID plan: 7h1nbzqjgwsp7, CHILD_NUMBER: 2
Check the value of SQL_ID and CHILD_NUMBER;
It could also be that the plan is no longer in the cursor cache (check v$ sql_plan)
NightWing wrote:
Randolf,
I don't understand. What you hear from the above statement that you mean A-lines and E will be incorrect, but the ratio between them remain the same. Therefore, you can deduct the bad things by comparing the differences.
Thus, A-lines always give a wrong result for cancellation of queries, isn't it?
Charlie,
I think that Martin gave a good explanation. Here's another example that hopefully makes more obvious things:
17:56:55 SQL >-things go very wrong here with a small buffer cache
17:56:55 SQL >-T2 lines are badly scattered when you access through T1. FK
17:56:55 SQL >--
17:56:55 SQL >-"Small job" approach would have been a good idea
17:56:55 SQL >-if the estimate of 100 iterations of the loop was correct!
17:56:55 SQL > select
17:56:55 (t2.attr2) count 2
17:56:55 3 of
17:56:55 4 t1
17:56:55 5, t2
17:56:55 6 where
17:56:55 7 /*------------------*/
17:56:55 8 trunc (t1.attr1) = 1
17:56:55 9 and trunc (t1.attr2) = 1
17:56:55 10 / *-* /.
17:56:55 11 and t1.fk = t2.id
17:56:55 12.
T1
*
ERROR on line 4:
ORA-01013: user has requested the cancellation of the current operation
Elapsed time: 00:04:58.30
18:01:53 SQL >
18:01:53 SQL > @xplan_extended_display_cursor ' ' ' ' 'ALLSTATS LAST + COST.
18:01:53 SQL > set echo off verify off termout off
SQL_ID, 353msax56jvvp, number of children 0
-------------------------------------
SELECT count (t2.attr2) from t1, t2 where
/ / *-* trunc (t1.attr1) = 1 and
trunc (T1.attr2) = 1 / *-* / and t1.fk = t2.id
Hash value of plan: 2900488714
------------------------------------------------------------------------------------------------------------------------------------
| ID | The NEST | DSB | Operation | Name | Begins | E - lines. Cost (% CPU). A - lines. A - time | Pads | Bed |
------------------------------------------------------------------------------------------------------------------------------------
| 0 | | 7. SELECT STATEMENT | | 1. | 4999 (100) | 0 | 00:00:00.01 | 0 | 0 |
| 1. 0 | 8 2 GLOBAL TRI | | 1. 1. | 0 | 00:00:00.01 | 0 | 0 |
| 2. 1. 5. NESTED LOOPS | | 1. | | 57516 | 00:04:58.26 | 173K | 30770 |
| 3. 2. 3. NESTED LOOPS | | 1. 100. 4999 (1) | 57516 | 00:00:21.06 | 116K | 3632.
|* 4 | 3. 1. TABLE ACCESS FULL | T1 | 1. 100. 4799 (1) | 57516 | 00:00:00.19 | 1008 | 1087 |
|* 5 | 3. 2. INDEX UNIQUE SCAN | T2_IDX | 57516 | 1. 1 (0) | 57516 | 00:00:20.82 | 115K | 2545 |
| 8 2 2 | 4. TABLE ACCESS BY INDEX ROWID | T2 | 57516 | 1. 2 (0) | 57516 | 00:04:37.14 | 57516 | 27138 |
------------------------------------------------------------------------------------------------------------------------------------
Information of predicates (identified by the operation identity card):
---------------------------------------------------
4 filter ((TRUNC ('T1'. "ATTR1") = 1 AND TRUNC ('T1'. " ATTR2') = 1))
5 - access("T1".") FK '= 'T2'.' (ID')
You say here that I canceled a query after about 5 minutes, and looking at the statistics of content (RowSource) I can already say the following:
1. the estimation of cardinality of T1 is far - the optimizer estimated 100 lines, but it actually generated more than 57000 lines when the query was cancelled. If this definitely seems like a candidate at the origin of the problems
2. the query has spent most of the time in search of random table T2
So while it is true that I don't know final A-lines of this cancelled query information, I can still say a lot of this and begin to deal with the problems identified so far.
Randolf
-
Star_transformation not shoiwng in the execution plan
Hello
We have a data warehouse and processing star not no projection in the execution plan.
STAR_TRANSFORMATION_ENABLED at the database level.
With the help of 11.2.0.3 on AIX
The fact table has index bitmap and fks to keys on tables dimesnion dimesnion.
Example query
Anthing we are ' % s '?select * from media m, retailer_transaction rt , retailer r where m.dimension_key = rt.plant_issue_id and rt.outlet_id = r.dimension_key and prod_num = 600
Found how interrogate more selective cause said to use star_transformation
added and plis_handled_year = 2013
and out_num 123423
Why is this good?
Is there a rule such as only highly selective queries will use transformation star?
Thank you
Published by: user5716448 on February 15, 2013 04:10Transformation of Star I it'sactually one expensive thing to do. So, the optimizer plans an ABC on this subject. This quote comes from a piece on the blog of optimizer Oracle [url https://blogs.oracle.com/optimizer/entry/star_transformation]:
"The transformation is performed based on cost - when the cost of the transformed plan is lower than that of the unprocessed plan. If dimension filters do not significantly reduce the amount of data to be extracted from the fact table, and then a full table scan is more effective. »
Cheers, APC
-
Same SQL different execution Plans in two different database instances.
Hi all
I run a different query on two instances of database. On one instance, the query takes 30 minutes and the other it takes 10 hours to complete. Data on the two instances are the same. When I generated the execution plan, on the two cases, it was different.
OS: Redhat Linux 5
Database version: 11.2.02 (on two instances)
Plan on 1 Instance that takes 30 minutes.
On the Instance 2SELECT X_REPLEN_RQST_INV_STG.BUSINESS_UNIT, NVL(SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', -1, 1) - LENGTH(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM)),'N/A') AS DOC_ID, W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_ITEM, W_PURCH_SCHEDULE_LINE_F.PURCH_SCHEDULE_NUM, X_REPLEN_RQST_INV_STG.INV_ITEM_ID, X_REPLEN_RQST_INV_STG.PROCESS_DATE AS REPLEN_PROCESS_DT, X_REPLEN_RQST_INV_STG.EXPECTED_DATE AS REPLEN_EXPECTED_DT, X_REPLEN_RQST_INV_STG.REORDER_QTY AS REPLEN_REORDER_QTY, X_ITEM_ATTRIB_D.REPLENISH_CLASS AS REPLEN_CLASS, X_REPLEN_RQST_INV_STG.REPLEN_STATUS, X_REPLEN_RQST_INV_STG.REPLENISH_TYPE AS REPLEN_TYPE, X_REPLEN_RQST_INV_STG.REQ_ID, X_REPLEN_RQST_INV_STG.REQ_LINE_NBR, X_REPLEN_RQST_INV_STG.REQ_SCHED_NBR, X_REPLEN_RQST_INV_STG.REQ_DISTRIB_NBR, P.FIRST_PO_DISP_DT, W_PURCH_SCHEDULE_LINE_F.DATASOURCE_NUM_ID FROM X_REPLEN_RQST_INV_STG LEFT OUTER JOIN W_RQSTN_LINE_COST_F ON X_REPLEN_RQST_INV_STG.RQSTN_LN_COST_INTG_ID = W_RQSTN_LINE_COST_F.INTEGRATION_ID AND X_REPLEN_RQST_INV_STG.DATASOURCE_NUM_ID = W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID LEFT OUTER JOIN W_PURCH_COST_F ON W_RQSTN_LINE_COST_F.INTEGRATION_ID = W_PURCH_COST_F.REQ_DISTR_INTG_ID AND W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID = W_PURCH_COST_F.DATASOURCE_NUM_ID INNER join W_PURCH_SCHEDULE_LINE_F ON W_PURCH_COST_F.PURCH_SCHEDULE_INTG_ID = W_PURCH_SCHEDULE_LINE_F.INTEGRATION_ID AND W_PURCH_COST_F.DATASOURCE_NUM_ID = W_PURCH_SCHEDULE_LINE_F.DATASOURCE_NUM_ID LEFT OUTER JOIN W_STATUS_D S1 ON W_RQSTN_LINE_COST_F.APPROVAL_STATUS_WID = S1.ROW_WID AND W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID = S1.DATASOURCE_NUM_ID AND 'PURCH_APPROVE' = S1.W_STATUS_CLASS LEFT OUTER JOIN W_STATUS_D S2 ON W_PURCH_COST_F.APPROVAL_STATUS_WID = S2.ROW_WID AND W_PURCH_COST_F.DATASOURCE_NUM_ID = S2.DATASOURCE_NUM_ID AND 'PURCH_APPROVE' = S2.W_STATUS_CLASS LEFT OUTER JOIN (SELECT p1.BUSINESS_UNIT, p1.PO_ID, MIN(p1.DATETIME_DISP) AS FIRST_PO_DISP_DT FROM X_PS_PO_DISPATCHED p1 GROUP BY p1.BUSINESS_UNIT, p1.PO_ID) P ON SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, 1, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', 1, 1)-1) = P.BUSINESS_UNIT AND SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', -1, 1) - LENGTH(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM)) = P.PO_ID LEFT OUTER JOIN W_PRODUCT_D ON W_PURCH_SCHEDULE_LINE_F.PRODUCT_WID = W_PRODUCT_D.ROW_WID LEFT OUTER JOIN X_ITEM_ATTRIB_D ON SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, 1, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', 1, 1)-1) = X_ITEM_ATTRIB_D.BUSINESS_UNIT AND W_PRODUCT_D.PART_NUM = X_ITEM_ATTRIB_D.INV_ITEM_ID WHERE TRIM(X_REPLEN_RQST_INV_STG.req_id) IS NOT NULL AND X_REPLEN_RQST_INV_STG.replenish_type = '1' AND X_REPLEN_RQST_INV_STG.replen_status = '4' AND X_ITEM_ATTRIB_D.REPLENISH_CLASS = 'A' AND S1.STATUS_CODE != 'X' AND S2.STATUS_CODE != 'X' AND P.FIRST_PO_DISP_DT IS NOT NULL Execution Plan ---------------------------------------------------------- Plan hash value: 1674299164 ---------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ---------------------------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | 461 | 13942 (1)| 00:04:11 | |* 1 | HASH JOIN | | 1 | 461 | 13942 (1)| 00:04:11 | | 2 | NESTED LOOPS | | | | | | | 3 | NESTED LOOPS | | 1 | 441 | 10409 (1)| 00:03:08 | | 4 | NESTED LOOPS | | 1 | 421 | 10407 (1)| 00:03:08 | | 5 | NESTED LOOPS | | 1 | 388 | 10406 (1)| 00:03:08 | | 6 | NESTED LOOPS | | 1 | 319 | 10405 (1)| 00:03:08 | | 7 | NESTED LOOPS | | 1 | 280 | 10404 (1)| 00:03:08 | |* 8 | HASH JOIN | | 1 | 220 | 10401 (1)| 00:03:08 | | 9 | NESTED LOOPS | | | | | | | 10 | NESTED LOOPS | | 16 | 2896 | 10392 (1)| 00:03:08 | |* 11 | TABLE ACCESS FULL | X_REPLEN_RQST_INV_STG | 16 | 2432 | 10375 (1)| 00:03:07 | |* 12 | INDEX UNIQUE SCAN | W_RQST_LN_CS_F_U1 | 1 | | 1 (0)| 00:00:01 | | 13 | TABLE ACCESS BY INDEX ROWID| W_RQSTN_LINE_COST_F | 1 | 29 | 2 (0)| 00:00:01 | |* 14 | TABLE ACCESS FULL | W_STATUS_D | 1 | 39 | 8 (0)| 00:00:01 | | 15 | TABLE ACCESS BY INDEX ROWID | W_PURCH_COST_F | 1 | 60 | 3 (0)| 00:00:01 | |* 16 | INDEX RANGE SCAN | W_PURCH_COST_F_M1 | 1 | | 2 (0)| 00:00:01 | |* 17 | TABLE ACCESS BY INDEX ROWID | W_STATUS_D | 1 | 39 | 1 (0)| 00:00:01 | |* 18 | INDEX UNIQUE SCAN | W_STATUS_D_P1 | 1 | | 0 (0)| 00:00:01 | | 19 | TABLE ACCESS BY INDEX ROWID | W_PURCH_SCHEDULE_LINE_F | 1 | 69 | 1 (0)| 00:00:01 | |* 20 | INDEX UNIQUE SCAN | W_PRCH_SC_LN_F_U1 | 1 | | 1 (0)| 00:00:01 | | 21 | TABLE ACCESS BY INDEX ROWID | W_PRODUCT_D | 1 | 33 | 1 (0)| 00:00:01 | |* 22 | INDEX UNIQUE SCAN | W_PRODUCT_D_P1 | 1 | | 0 (0)| 00:00:01 | |* 23 | INDEX RANGE SCAN | X_ITEM_ATTRIB_U01 | 1 | | 1 (0)| 00:00:01 | |* 24 | TABLE ACCESS BY INDEX ROWID | X_ITEM_ATTRIB_D | 1 | 20 | 2 (0)| 00:00:01 | | 25 | VIEW | | 1428K| 27M| 3529 (2)| 00:01:04 | |* 26 | FILTER | | | | | | | 27 | HASH GROUP BY | | 1428K| 27M| 3529 (2)| 00:01:04 | | 28 | TABLE ACCESS FULL | X_PS_PO_DISPATCHED | 1428K| 27M| 3493 (1)| 00:01:03 | ----------------------------------------------------------------------------------------------------------------
In the second execution plan, we can see that it uses a Cartesian product.SELECT X_REPLEN_RQST_INV_STG.BUSINESS_UNIT, NVL(SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', -1, 1) - LENGTH(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM)),'N/A') AS DOC_ID, W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_ITEM, W_PURCH_SCHEDULE_LINE_F.PURCH_SCHEDULE_NUM, X_REPLEN_RQST_INV_STG.INV_ITEM_ID, X_REPLEN_RQST_INV_STG.PROCESS_DATE AS REPLEN_PROCESS_DT, X_REPLEN_RQST_INV_STG.EXPECTED_DATE AS REPLEN_EXPECTED_DT, X_REPLEN_RQST_INV_STG.REORDER_QTY AS REPLEN_REORDER_QTY, X_ITEM_ATTRIB_D.REPLENISH_CLASS AS REPLEN_CLASS, X_REPLEN_RQST_INV_STG.REPLEN_STATUS, X_REPLEN_RQST_INV_STG.REPLENISH_TYPE AS REPLEN_TYPE, X_REPLEN_RQST_INV_STG.REQ_ID, X_REPLEN_RQST_INV_STG.REQ_LINE_NBR, X_REPLEN_RQST_INV_STG.REQ_SCHED_NBR, X_REPLEN_RQST_INV_STG.REQ_DISTRIB_NBR, P.FIRST_PO_DISP_DT, W_PURCH_SCHEDULE_LINE_F.DATASOURCE_NUM_ID FROM X_REPLEN_RQST_INV_STG LEFT OUTER JOIN W_RQSTN_LINE_COST_F ON X_REPLEN_RQST_INV_STG.RQSTN_LN_COST_INTG_ID = W_RQSTN_LINE_COST_F.INTEGRATION_ID AND X_REPLEN_RQST_INV_STG.DATASOURCE_NUM_ID = W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID LEFT OUTER JOIN W_PURCH_COST_F ON W_RQSTN_LINE_COST_F.INTEGRATION_ID = W_PURCH_COST_F.REQ_DISTR_INTG_ID AND W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID = W_PURCH_COST_F.DATASOURCE_NUM_ID INNER join W_PURCH_SCHEDULE_LINE_F ON W_PURCH_COST_F.PURCH_SCHEDULE_INTG_ID = W_PURCH_SCHEDULE_LINE_F.INTEGRATION_ID AND W_PURCH_COST_F.DATASOURCE_NUM_ID = W_PURCH_SCHEDULE_LINE_F.DATASOURCE_NUM_ID LEFT OUTER JOIN W_STATUS_D S1 ON W_RQSTN_LINE_COST_F.APPROVAL_STATUS_WID = S1.ROW_WID AND W_RQSTN_LINE_COST_F.DATASOURCE_NUM_ID = S1.DATASOURCE_NUM_ID AND 'PURCH_APPROVE' = S1.W_STATUS_CLASS LEFT OUTER JOIN W_STATUS_D S2 ON W_PURCH_COST_F.APPROVAL_STATUS_WID = S2.ROW_WID AND W_PURCH_COST_F.DATASOURCE_NUM_ID = S2.DATASOURCE_NUM_ID AND 'PURCH_APPROVE' = S2.W_STATUS_CLASS LEFT OUTER JOIN (SELECT p1.BUSINESS_UNIT, p1.PO_ID, MIN(p1.DATETIME_DISP) AS FIRST_PO_DISP_DT FROM X_PS_PO_DISPATCHED p1 GROUP BY p1.BUSINESS_UNIT, p1.PO_ID) P ON SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, 1, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', 1, 1)-1) = P.BUSINESS_UNIT AND SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', -1, 1) - LENGTH(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM)) = P.PO_ID LEFT OUTER JOIN W_PRODUCT_D ON W_PURCH_SCHEDULE_LINE_F.PRODUCT_WID = W_PRODUCT_D.ROW_WID LEFT OUTER JOIN X_ITEM_ATTRIB_D ON SUBSTR (W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, 1, INSTR(W_PURCH_SCHEDULE_LINE_F.PURCH_ORDER_NUM, '~', 1, 1)-1) = X_ITEM_ATTRIB_D.BUSINESS_UNIT AND W_PRODUCT_D.PART_NUM = X_ITEM_ATTRIB_D.INV_ITEM_ID WHERE TRIM(X_REPLEN_RQST_INV_STG.req_id) IS NOT NULL AND X_REPLEN_RQST_INV_STG.replenish_type = '1' AND X_REPLEN_RQST_INV_STG.replen_status = '4' AND X_ITEM_ATTRIB_D.REPLENISH_CLASS = 'A' AND S1.STATUS_CODE != 'X' AND S2.STATUS_CODE != 'X' AND P.FIRST_PO_DISP_DT IS NOT NULL --------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | --------------------------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | 420 | 19630 (1)| 00:04:35 | | 1 | NESTED LOOPS | | | | | | | 2 | NESTED LOOPS | | 1 | 420 | 19630 (1)| 00:04:35 | | 3 | NESTED LOOPS | | 1 | 400 | 19627 (1)| 00:04:35 | | 4 | NESTED LOOPS | | 1 | 386 | 19626 (1)| 00:04:35 | | 5 | NESTED LOOPS | | 1 | 317 | 19625 (1)| 00:04:35 | | 6 | NESTED LOOPS | | 1 | 289 | 19624 (1)| 00:04:35 | | 7 | NESTED LOOPS | | 1 | 229 | 19621 (1)| 00:04:35 | | 8 | NESTED LOOPS | | 1 | 201 | 19620 (1)| 00:04:35 | | 9 | MERGE JOIN CARTESIAN | | 1 | 172 | 19619 (1)| 00:04:35 | | 10 | VIEW | | 1 | 20 | 4942 (2)| 00:01:10 | |* 11 | FILTER | | | | | | | 12 | HASH GROUP BY | | 1 | 20 | 4942 (2)| 00:01:10 | | 13 | TABLE ACCESS FULL | X_PS_PO_DISPATCHED | 1428K| 27M| 4902 (1)| 00:01:09 | | 14 | BUFFER SORT | | 1 | 152 | 19619 (1)| 00:04:35 | |* 15 | TABLE ACCESS FULL | X_REPLEN_RQST_INV_STG | 1 | 152 | 14676 (1)| 00:03:26 | | 16 | TABLE ACCESS BY INDEX ROWID| W_RQSTN_LINE_COST_F | 1 | 29 | 1 (0)| 00:00:01 | |* 17 | INDEX UNIQUE SCAN | W_RQST_LN_CS_F_U1 | 1 | | 1 (0)| 00:00:01 | |* 18 | TABLE ACCESS BY INDEX ROWID | W_STATUS_D | 1 | 28 | 1 (0)| 00:00:01 | |* 19 | INDEX UNIQUE SCAN | W_STATUS_D_P1 | 1 | | 0 (0)| 00:00:01 | | 20 | TABLE ACCESS BY INDEX ROWID | W_PURCH_COST_F | 3 | 180 | 3 (0)| 00:00:01 | |* 21 | INDEX RANGE SCAN | W_PURCH_COST_F_M1 | 1 | | 2 (0)| 00:00:01 | |* 22 | TABLE ACCESS BY INDEX ROWID | W_STATUS_D | 1 | 28 | 1 (0)| 00:00:01 | |* 23 | INDEX UNIQUE SCAN | W_STATUS_D_P1 | 1 | | 0 (0)| 00:00:01 | |* 24 | TABLE ACCESS BY INDEX ROWID | W_PURCH_SCHEDULE_LINE_F | 1 | 69 | 1 (0)| 00:00:01 | |* 25 | INDEX UNIQUE SCAN | W_PRCH_SC_LN_F_U1 | 1 | | 1 (0)| 00:00:01 | |* 26 | TABLE ACCESS BY INDEX ROWID | W_PRODUCT_D | 1 | 14 | 1 (0)| 00:00:01 | |* 27 | INDEX UNIQUE SCAN | W_PRODUCT_D_P1 | 1 | | 0 (0)| 00:00:01 | |* 28 | INDEX RANGE SCAN | X_ITEM_ATTRIB_U01 | 1 | | 2 (0)| 00:00:01 | |* 29 | TABLE ACCESS BY INDEX ROWID | X_ITEM_ATTRIB_D | 1 | 20 | 3 (0)| 00:00:01 | ---------------------------------------------------------------------------------------------------------------
Thank you.Oceaner wrote:
Hi allI run a different query on two instances of database. On one instance, the query takes 30 minutes and the other it takes 10 hours to complete. Data on the two instances are the same. When I generated the execution plan, on the two cases, it was different.
The most obvious difference is the subquery total inline:
Quick plan:
---------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ---------------------------------------------------------------------------------------------------------------- |* 1 | HASH JOIN | | 1 | 461 | 13942 (1)| 00:04:11 | | 2 | NESTED LOOPS | | | | | | ... | 25 | VIEW | | 1428K| 27M| 3529 (2)| 00:01:04 | |* 26 | FILTER | | | | | | | 27 | HASH GROUP BY | | 1428K| 27M| 3529 (2)| 00:01:04 | | 28 | TABLE ACCESS FULL | X_PS_PO_DISPATCHED | 1428K| 27M| 3493 (1)| 00:01:03 | ----------------------------------------------------------------------------------------------------------------Plan of slow
--------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | --------------------------------------------------------------------------------------------------------------- ... | 9 | MERGE JOIN CARTESIAN | | 1 | 172 | 19619 (1)| 00:04:35 | | 10 | VIEW | | 1 | 20 | 4942 (2)| 00:01:10 | |* 11 | FILTER | | | | | | | 12 | HASH GROUP BY | | 1 | 20 | 4942 (2)| 00:01:10 | | 13 | TABLE ACCESS FULL | X_PS_PO_DISPATCHED | 1428K| 27M| 4902 (1)| 00:01:09 | | 14 | BUFFER SORT | | 1 | 152 | 19619 (1)| 00:04:35 | |* 15 | TABLE ACCESS FULL | X_REPLEN_RQST_INV_STG | 1 | 152 | 14676 (1)| 00:03:26 | ... ---------------------------------------------------------------------------------------------------------------Note the value of lines given to the operation - VIEW
One for the slow plan - that of why Oracle chose display online as a starting point for the query and used a Cartesian merge join to join to the table nearby.1.4Million for the fast plan - this is why Oracle has decided to visit the table set and use it as the table of the probe in a hash join.
I think we can assume that the estimation of the single line is a big mistake. We can also assume that the two series of statistics (or possibly index definitions) on the tables two versions do not match.
It would be useful, of course, see the comprehensive implementation plan, and preferably one from memory: (http://jonathanlewis.wordpress.com/2006/12/22/dbms_xplan-again/).
You might notice that you outer joins are not relevant - maybe this SQL is a framework for a more general query - given that the plan shows that they have all been turned away.
Concerning
Jonathan Lewis
http://jonathanlewis.WordPress.com
Author: core Oracle -
With concatenation execution plan
Hi all
Could someone help to find out why concatenation is used by the optimizer and how I avoid it.
Oracle Version: 10.2.0.4
table of IFE.entityids were brought - partitioned on the data_provider column.select * from ( select distinct EntityType, EntityID, DateModified, DateCreated, IsDeleted from ife.EntityIDs i join (select orgid from equifaxnormalize.org_relationships where orgid is not null and related_orgid is not null and ((Date_Modified >= to_date('2011-06-12 14:00:00','yyyy-mm-dd hh24:mi:ss') and Date_Modified < to_date('2011-06-13 14:00:00','yyyy-mm-dd hh24:mi:ss')) OR (Date_Created >= to_date('2011-06-12 14:00:00','yyyy-mm-dd hh24:mi:ss') and Date_Created < to_date('2011-06-13 14:00:00','yyyy-mm-dd hh24:mi:ss')) ) ) r on(r.orgid= i.entityid) where EntityType = 1 and ((DateModified >= to_date('2011-06-12 14:00:00','yyyy-mm-dd hh24:mi:ss') and DateModified < to_date('2011-06-13 14:00:00','yyyy-mm-dd hh24:mi:ss')) OR (DateCreated >= to_date('2011-06-12 14:00:00','yyyy-mm-dd hh24:mi:ss') and DateCreated < to_date('2011-06-13 14:00:00','yyyy-mm-dd hh24:mi:ss')) ) and ( IsDeleted = 0) and IsDistributable = 1 and EntityID >= 0 order by EntityID --order by NLSSORT(EntityID,'NLS_SORT=BINARY') ) where rownum <= 10; Execution Plan ---------------------------------------------------------- Plan hash value: 227906424 ------------------------------------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop | ------------------------------------------------------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 10 | 570 | 39 (6)| 00:00:01 | | | |* 1 | COUNT STOPKEY | | | | | | | | | 2 | VIEW | | 56 | 3192 | 39 (6)| 00:00:01 | | | |* 3 | SORT ORDER BY STOPKEY | | 56 | 3416 | 39 (6)| 00:00:01 | | | | 4 | HASH UNIQUE | | 56 | 3416 | 38 (3)| 00:00:01 | | | | 5 | CONCATENATION | | | | | | | | |* 6 | TABLE ACCESS BY INDEX ROWID | ORG_RELATIONSHIPS | 1 | 29 | 1 (0)| 00:00:01 | | | | 7 | NESTED LOOPS | | 27 | 1647 | 17 (0)| 00:00:01 | | | | 8 | TABLE ACCESS BY GLOBAL INDEX ROWID| ENTITYIDS | 27 | 864 | 4 (0)| 00:00:01 | ROWID | ROWID | |* 9 | INDEX RANGE SCAN | UX_TYPE_MOD_DIST_DEL_ENTITYID | 27 | | 2 (0)| 00:00:01 | | | |* 10 | INDEX RANGE SCAN | IX_EFX_ORGRELATION_ORGID | 1 | | 1 (0)| 00:00:01 | | | |* 11 | TABLE ACCESS BY INDEX ROWID | ORG_RELATIONSHIPS | 1 | 29 | 1 (0)| 00:00:01 | | | | 12 | NESTED LOOPS | | 29 | 1769 | 20 (0)| 00:00:01 | | | | 13 | PARTITION RANGE ALL | | 29 | 928 | 5 (0)| 00:00:01 | 1 | 3 | |* 14 | TABLE ACCESS BY LOCAL INDEX ROWID| ENTITYIDS | 29 | 928 | 5 (0)| 00:00:01 | 1 | 3 | |* 15 | INDEX RANGE SCAN | IDX_ENTITYIDS_ETYPE_DC | 29 | | 4 (0)| 00:00:01 | 1 | 3 | |* 16 | INDEX RANGE SCAN | IX_EFX_ORGRELATION_ORGID | 1 | | 1 (0)| 00:00:01 | | | ------------------------------------------------------------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 1 - filter(ROWNUM<=10) 3 - filter(ROWNUM<=10) 6 - filter(("DATE_MODIFIED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "DATE_MODIFIED"<TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss') OR "DATE_CREATED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "DATE_CREATED"<TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss')) AND "RELATED_ORGID" IS NOT NULL) 9 - access("I"."ENTITYTYPE"=1 AND "I"."DATEMODIFIED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "I"."ISDISTRIBUTABLE"=1 AND "I"."ISDELETED"=0 AND "I"."ENTITYID">=0 AND "I"."DATEMODIFIED"<=TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss')) filter("I"."ISDISTRIBUTABLE"=1 AND "I"."ISDELETED"=0 AND "I"."ENTITYID">=0) 10 - access("ORGID"="I"."ENTITYID") filter("ORGID" IS NOT NULL AND "ORGID">=0) 11 - filter(("DATE_MODIFIED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "DATE_MODIFIED"<TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss') OR "DATE_CREATED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "DATE_CREATED"<TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss')) AND "RELATED_ORGID" IS NOT NULL) 14 - filter("I"."ISDISTRIBUTABLE"=1 AND "I"."ISDELETED"=0 AND (LNNVL("I"."DATEMODIFIED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss')) OR LNNVL("I"."DATEMODIFIED"<=TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss'))) AND "I"."ENTITYID">=0) 15 - access("I"."ENTITYTYPE"=1 AND "I"."DATECREATED">=TO_DATE(' 2011-06-12 14:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "I"."DATECREATED"<=TO_DATE(' 2011-06-13 14:00:00', 'syyyy-mm-dd hh24:mi:ss')) 16 - access("ORGID"="I"."ENTITYID") filter("ORGID" IS NOT NULL AND "ORGID">=0)
Is there a better way to rewrite this sql OR is it possible to eliminate the concatenation?
Thank youAs general approach, see the tuning wires:
[url http://forums.oracle.com/forums/thread.jspa?threadID=863295] How to post a sql tuning request
[url http://forums.oracle.com/forums/thread.jspa?messageID=1812597] When your query takes too longThe approach is essentially, for where the estimates are more inaccurate compared to actual expenditures and can find the reasons why.
Most of the time, if the statistics are accurate and then the estimates are correct, you will have a good plan. -
Hello
can you be kind to such me which is better and why (based on the columns of the execution Plan):
AND:Execution Plan A ---------------------------------------------------------- 0 SELECT STATEMENT Optimizer=CHOOSE (Cost=2 Card=1 Bytes=17) 1 0 SORT (AGGREGATE) 2 1 INDEX (RANGE SCAN) OF 'TEST_IDX' (NON-UNIQUE) (Cost=2 Card=1 Bytes=17) 3 2 SORT (AGGREGATE) 4 3 FIRST ROW (Cost=2 Card=6 Bytes=60) 5 4 INDEX (RANGE SCAN (MIN/MAX)) OF 'TEST_IDX' (NON-UNIQUE) (Cost=2 Card=1060) tkprof gek1_ora_16520.trc gek1_ora_16520.out explain=scott/tiger sort=exeela sys=no call count cpu elapsed disk query current rows ------- ------ -------- ---------- ---------- ---------- ---------- ---------- Parse 1 0.00 0.00 0 0 0 0 Execute 1 0.00 0.00 0 2 0 0 Fetch 2 0.00 0.00 0 2 0 1 ------- ------ -------- ---------- ---------- ---------- ---------- ---------- total 4 0.00 0.00 0 4 0 1
Thank you.Execution Plan B ---------------------------------------------------------- 0 SELECT STATEMENT Optimizer=CHOOSE (Cost=2 Card=1 Bytes=13) 1 0 COUNT (STOPKEY) 2 1 VIEW (Cost=2 Card=6 Bytes=78) 3 2 INDEX (RANGE SCAN DESCENDING) OF 'TEST_IDX' (NON-UNIQUE) (Cost=2 Card=6 Bytes=102) tkprof gek1_ora_16521.trc gek1_ora_16521.out explain=scott/tiger sort=exeela sys=no call count cpu elapsed disk query current rows ------- ------ -------- ---------- ---------- ---------- ---------- ---------- Parse 1 0.03 0.06 2 41 0 0 Execute 1 0.00 0.00 0 0 0 0 Fetch 2 0.00 0.00 0 2 0 1 ------- ------ -------- ---------- ---------- ---------- ---------- ---------- total 4 0.03 0.06 2 43 0 1What is the version of db? Why you have not posted the actual queries as well? In a very generic view, the first is more beautiful, but is purely based on the elapsed time. What you are actually looking for?
HTH
Aman... -
The execution plan changes for the same query.
Hi all
This issue was raised before also, but still not able to find the real cause of this.
Thread1:
Thread2:
It comes, sometimes hammers server 100% CPU utilization with free latch and buffer busy wait events.
We found a single query consumes high CPU usage that is run by different sessions.
This query have two types of execution plans, where one is accurate and is not (its primary key hit index index no appropriate means present on the table)
Because its primary key index hit repeatedly at various sessions, some sessions are powerful db file sequential read and a few sessions waiting buffer busy waits for event. Also during this time a few sessions waiting for latch free event.
My doubt is how to sql even with different literal values execution plan changes and causes a prob.
How to avoid this... why its different execution plan using (I mean bad index PK)select count(*),event from v$session_wait group by event; COUNT(*) EVENT ---------- ---------------------------------------------------------------- 165 SQL*Net message from client 1 SQL*Net message to client 3 buffer busy waits 2 db file parallel read 18 db file sequential read 10 latch free 5 log file sync 1 pmon timer 6 rdbms ipc message 1 smon timer SQL> select sid from v$session_wait where event='db file sequential read'; SID ---------- 26 58 82 107 116 223 212 203 192 173 161 157 150 147 254 238 229 112 101 81 68 SQL> select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID; Enter value for sid: 161 old 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID new 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=161 SPID SID SERIAL# PROGRAM --------- ---------- ---------- ------------------------------------------------ 4231 161 49569 oracle@tfrdb3 (TNS V1-V3) SQL> select sql_text from v$process a, v$session b, 2 3 v$sql c where a.addr = b.paddr and b.sql_hash_value = c.hash_value and a.spid = &PID; 4 5 6 7 Enter value for pid: 4231 old 7: a.spid = &PID new 7: a.spid = 4231 SQL_TEXT -------------------------------------------------------------------------------- SELECT ERROR,TIME_STAMP,O_RESOURCE,QUEUE,NEW_QUEUE FROM LOG WHERE ID = '09292AMR 10B41FE' AND TYPE IN (11, 28, 25, 18, 60, 13) AND (LOG_SEQ>'234225222' OR TYPE = 18 AND LOG_SEQ='234225222') ORDER BY TIME_STAMP ASC SQL> set autotrace traceonly exp SQL> SELECT ERROR,TIME_STAMP,O_RESOURCE,QUEUE,NEW_QUEUE FROM amrwf1.LOG WHERE ID = '09292AMR10B41FE' AND TYPE IN (11, 28, 25, 18, 60, 13) AND (LOG_SEQ>'234225222' OR TYPE =18 AND LOG_SEQ='234225222') ORDER BY TIME_STAMP ASC; Execution Plan ---------------------------------------------------------- 0 SELECT STATEMENT Optimizer=CHOOSE (Cost=11 Card=2 Bytes=126) 1 0 SORT (ORDER BY) (Cost=11 Card=2 Bytes=126) 2 1 CONCATENATION 3 2 TABLE ACCESS (BY INDEX ROWID) OF 'LOG' (Cost=4 Card=1 Bytes=63) 4 3 INDEX (UNIQUE SCAN) OF 'PK_LOG_LOG_SEQ' (UNIQUE) (Co st=3 Card=1) 5 2 TABLE ACCESS (BY INDEX ROWID) OF 'LOG' (Cost=4 Card=1 Bytes=63) 6 5 INDEX (RANGE SCAN) OF 'PK_LOG_LOG_SEQ' (UNIQUE) (Cos t=3 Card=1) SQL> select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID; Enter value for sid: 147 old 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID new 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=147 SPID SID SERIAL# PROGRAM --------- ---------- ---------- ------------------------------------------------ 6255 147 38306 oracle@tfrdb3 (TNS V1-V3) SQL> select sql_text from v$process a, v$session b, v$sql c 2 3 where a.addr = b.paddr and b.sql_hash_value = c.hash_value and a.spid = &PID; 4 5 6 7 Enter value for pid: 6255 old 7: a.spid = &PID new 7: a.spid = 6255 SQL_TEXT -------------------------------------------------------------------------------- SELECT ERROR,TIME_STAMP,O_RESOURCE,QUEUE,NEW_QUEUE FROM LOG WHERE ID = '09273AMR 62B4894' AND TYPE IN (11, 28, 25, 18, 60, 13) AND (LOG_SEQ>'223324996' OR TYPE = 18 AND LOG_SEQ='223324996') ORDER BY TIME_STAMP ASC SQL> set autotrace traceonly exp SQL> SELECT ERROR,TIME_STAMP,O_RESOURCE,QUEUE,NEW_QUEUE FROM amrwf1.LOG WHERE ID = '09273AMR62B4894' AND TYPE IN (11, 28, 25, 18, 60, 13) AND (LOG_SEQ>'223324996' OR TYPE =18 AND LOG_SEQ='223324996') ORDER BY TIME_STAMP ASC; Execution Plan ---------------------------------------------------------- 0 SELECT STATEMENT Optimizer=CHOOSE (Cost=1538 Card=736 Bytes= 46368) 1 0 SORT (ORDER BY) (Cost=1538 Card=736 Bytes=46368) 2 1 TABLE ACCESS (BY INDEX ROWID) OF 'LOG' (Cost=1527 Card=7 36 Bytes=46368) 3 2 INDEX (RANGE SCAN) OF 'LOG_ID' (NON-UNIQUE) (Cost=32 C ard=736) SQL> select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID; Enter value for sid: 82 old 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=&SID new 1: select spid, sid, s.serial#, p.program from v$session s, v$process p where paddr=addr and sid=82 SPID SID SERIAL# PROGRAM --------- ---------- ---------- ------------------------------------------------ 6172 82 45378 oracle@tfrdb3 (TNS V1-V3) SQL> select sql_text from v$process a, v$session b, v$sql c where a.addr = b.paddr and b.sql_hash_value = c.hash_value and 2 3 a.spid = &PID; 4 5 6 7 Enter value for pid: 6172 old 7: a.spid = &PID new 7: a.spid = 6172 SQL_TEXT -------------------------------------------------------------------------------- INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) SQL_TEXT -------------------------------------------------------------------------------- INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V SQL_TEXT -------------------------------------------------------------------------------- 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010) INSERT INTO LOG (ID,TIME_STAMP,TYPE,ERROR,INSTANCE,RULE_NUM,RULE_TYPE,PRIORITY,F LAGS,NAME,BATCH,O_RESOURCE,QUEUE,NEW_QUEUE,SERVER,FORM,WORKSET) VALUES (:V001,:V 002,11,0,0,3,1,0,1,:V003,:V004,:V005,:V006,:V007,:V008,:V009,:V010)
Is it possible to avoid this?
If any details please check out some of my previous post on this specific URL (above)
-YasserMy doubt is how to sql even with different literal values execution plan changes and causes a prob.
Different literal values cause analysis difficult.
Hard analysis includes the re-evaluation of the best path.
Literal value is included in the assessment of the selectivity for the scan interval (log_seq >...)See
http://www.centrexcc.com/A%20Look%20under%20The%20Hood%20Of%20CBO%20-%20The%2010053%20Event.ppt.PDF
http://www.centrexcc.com/fallacies%20Of%20The%20Cost%20Based%20Optimizer.PDF
more the book of Jonathan Lewis which other threads, I believe that you already have.You must lower your CPU.
Previous discussions, if the situation is still the same, it sounded like hard analysis particularly with this SELECTION against the NEWSPAPER plays an important role in that.How to avoid this... why its different execution plan using (I mean bad index PK)
The points raised in the previous discussion remain valid.
-Do you have access to this SQL to change?
for example using bind variable or trick it if necessary due to problems caused by data as discussed in the previous thread.
- Or you could it repoint the view to a view and a hint?
-If a particular user makes this sql, could affect you cursor_sharing just for this user. If not, you should consider implementing pan-Canadian database.Oracle 8.1.6 still?
-
Hi all
Can I get a query to see if there are several implementation plans for a query.
I tried with v$ sql where CHILD_NUMBER can be 0 or 1 at a time.
What interests me is any plans for a particular sql.
Also I can get good technical quality for outln paper / and the profile of sql. I have some links but do not have enough to her.
Thank you
Rana.user582224 wrote:
Can I get a query to see if there are several implementation plans for a query.
I tried with v$ sql where CHILD_NUMBER can be 0 or 1 at a time.What interests me is any plans for a particular sql.
Rana,
If you are interested in what you have now in the pool that is shared, you can start with V$ SQL, V$ SQLAREA, SQLSTATS $ V and V$ SQL_PLAN.
Note that different the same statement execution plans may already have aged out of the shared pool, that's why this information does not necessarily reflect the number of versions you may have had over time.
If you are already on 10g or later, you can use the DBMS_XPLAN. Function DISPLAY_CURSOR. If you specify "NULL" for the second parameter ("cursor_child_no"), it will show you all the sliders of the specified SQL_ID child execution plans.
If you have 10g or later and a CWA license, you can use DBA_HIST_SQL_PLAN and DBMS_XPLAN. DISPLAY_AWR to get stored for a particular SQL_ID. Note different execution plans that AWR cannot capture your particular statement, since it only samples consumers albums according to certain thresholds.
If you do not have a license of AWR you will always get a similar historical view of your plans with using STATSPACK snapshot level > = 6 that captures SQL execution plans. Once your statement again could not enjoy according to the load and thresholds (STATSPACK documentation: "to collect plans for all statements in the shared pool, you can temporarily specify the threshold of executions (i_executions_th) to be zero (0) for these snapshots").
Kind regards
RandolfOracle related blog stuff:
http://Oracle-Randolf.blogspot.com/SQLTools ++ for Oracle (Open source Oracle GUI for Windows):
http://www.sqltools-plusplus.org:7676 /.
http://sourceforge.NET/projects/SQLT-pp/ -
CBO does not choose the right execution plan
Database: Oracle 9.2.0.6 EA
OS: Solaris 9
I'm trying to settle a query that is generated through Siebel Analytics. I see a behavior which is confusing to me but I hope, would be 'basic' for someone like JPL.
The query is based on a total of 7 tables. If I comment on the 2 dimension tables, the query takes the right index finger on the fact table. However, at the point where I have to add another table to the query, the plan goes awry.
The query with 5 tables is as below:
And the implementation plan is as below:select count(distinct decode( T30256.HEADER_FLG , 'N' , T30256.ROW_WID ) ) as c1, T352305.DAY_DT as c2, case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end as c3, T352305.ROW_WID as c5 from W_PRODUCT_D T30955, W_PRDATTRNM_D T44643, W_DAY_D T352305, W_ORDERITEM_F T30256, W_PRDATTR_D T40081 where ( T30955.ROW_WID = T44643.ROW_WID and T30256.LAST_UPD_DT_WID = T352305.ROW_WID and T30256.PROD_ATTRIB_WID = T40081.ROW_WID and T30256.PROD_WID = T30955.ROW_WID and T30955.PROD_NAME = 'Mobile Subscription' and (case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end in ('BT150BB-18M', 'BT250BB-18M', 'BT50BB-18M', 'BT600BB-18M')) and T352305.DAY_DT between TO_DATE('2008-09-27' , 'YYYY-MM-DD') - 7 and TO_DATE('2008-09-27' , 'YYYY-MM-DD') - 1 ) group by T352305.ROW_WID, T352305.DAY_DT, case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end ;
Note how the dimension tables W_PRODUCT_D & W_DAY_D are connected by a Cartesian join before joining the fact W_ORDERITEM_F table using the composite index 'W_ORDERITEM_F_INDX9 '. This index is composed of LAST_UPD_DT_WID, PROD_WID and ACTION_TYPE_WID, which are foreign keys to the dimension tables.---------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| ---------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 269 | 25824 | 18660 (3)| | 1 | SORT GROUP BY | | 269 | 25824 | 18660 (3)| | 2 | NESTED LOOPS | | 269 | 25824 | 18658 (3)| | 3 | NESTED LOOPS | | 6826 | 579K| 4734 (3)| | 4 | MERGE JOIN CARTESIAN | | 8 | 544 | 6 (17)| | 5 | NESTED LOOPS | | 1 | 54 | 4 (25)| | 6 | TABLE ACCESS BY INDEX ROWID| W_PRODUCT_D | 1 | 37 | 3 (34)| |* 7 | INDEX RANGE SCAN | W_PRODUCT_D_M2 | 1 | | 2 (50)| | 8 | TABLE ACCESS BY INDEX ROWID| W_PRDATTRNM_D | 1 | 17 | 2 (50)| |* 9 | INDEX UNIQUE SCAN | W_PRDATTRNM_D_P1 | 1 | | | | 10 | BUFFER SORT | | 8 | 112 | 4 (0)| | 11 | TABLE ACCESS BY INDEX ROWID| W_DAY_D | 8 | 112 | 3 (34)| |* 12 | INDEX RANGE SCAN | W_DAY_D_M39 | 8 | | 2 (50)| | 13 | TABLE ACCESS BY INDEX ROWID | W_ORDERITEM_F | 849 | 16131 | 592 (3)| |* 14 | INDEX RANGE SCAN | W_ORDERITEM_F_INDX9 | 852 | | 4 (25)| |* 15 | INDEX RANGE SCAN | W_PRDATTR_D_M29_T1 | 1 | 9 | 3 (34)| ----------------------------------------------------------------------------------------------
Now, if I add several tables to the query:
I added a single table W_ORDER_D to the query and execution plan is:select count(distinct decode( T30256.HEADER_FLG , 'N' , T30256.ROW_WID ) ) as c1, T352305.DAY_DT as c2, case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end as c3, T30371.X_BT_DLR_GROUP as c4, T352305.ROW_WID as c5 from W_PRODUCT_D T30955, W_PRDATTRNM_D T44643, W_DAY_D T352305, W_ORDERITEM_F T30256, W_ORDER_D T30371, W_PRDATTR_D T40081 where ( T30955.ROW_WID = T44643.ROW_WID and T30256.LAST_UPD_DT_WID = T352305.ROW_WID and T30256.PROD_ATTRIB_WID = T40081.ROW_WID and T30256.PROD_WID = T30955.ROW_WID and T30256.ORDER_WID = T30371.ROW_WID and T30955.PROD_NAME = 'Mobile Subscription' and T30371.STATUS_CD = 'Complete' and T30371.ORDER_TYPE = 'Sales Order' and (case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end in ('BT150BB-18M', 'BT250BB-18M', 'BT50BB-18M', 'BT600BB-18M')) and T352305.DAY_DT between TO_DATE('2008-09-27' , 'YYYY-MM-DD') - 7 and TO_DATE('2008-09-27' , 'YYYY-MM-DD') - 1 ) group by T30371.X_BT_DLR_GROUP, T352305.ROW_WID, T352305.DAY_DT, case when T44643.PRODUCT_CLASS_NAME = 'MobileSubscription' then T40081.ATTR15_CHAR_VAL else 'Unspecified' end;
Now, CBO choose the composite index and the cost has also increased to 78695. But if I just add a / * + SORTED * / referring to the above query, whereas it should join the dimension tables prior to joining the fact table, then the cost falls to 20913. This means that CBO is not choose the plan with the lowest cost. I tried increasing the optimizer_max_permutations 80000, setting session optimizer_dynamic_sampling level 8 (just to see if it works), but without success.----------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| ----------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 44 | 6336 | 78695 (3)| | 1 | SORT GROUP BY | | 44 | 6336 | 78695 (3)| | 2 | NESTED LOOPS | | 44 | 6336 | 78694 (3)| | 3 | NESTED LOOPS | | 269 | 27707 | 78145 (3)| |* 4 | HASH JOIN | | 6826 | 626K| 64221 (3)| | 5 | TABLE ACCESS BY INDEX ROWID | W_DAY_D | 8 | 112 | 4 (25)| |* 6 | INDEX RANGE SCAN | W_DAY_D_M39 | 1 | | 3 (34)| | 7 | TABLE ACCESS BY INDEX ROWID | W_ORDERITEM_F | 86886 | 2206K| 64197 (3)| | 8 | NESTED LOOPS | | 87004 | 6797K| 64200 (3)| | 9 | NESTED LOOPS | | 1 | 54 | 4 (25)| | 10 | TABLE ACCESS BY INDEX ROWID| W_PRODUCT_D | 1 | 37 | 3 (34)| |* 11 | INDEX RANGE SCAN | W_PRODUCT_D_M2 | 1 | | 2 (50)| | 12 | TABLE ACCESS BY INDEX ROWID| W_PRDATTRNM_D | 1 | 17 | 2 (50)| |* 13 | INDEX UNIQUE SCAN | W_PRDATTRNM_D_P1 | 1 | | | |* 14 | INDEX RANGE SCAN | W_ORDERITEM_F_N6 | 86886 | | 212 (18)| |* 15 | INDEX RANGE SCAN | W_PRDATTR_D_M29_T1 | 1 | 9 | 3 (34)| |* 16 | INDEX RANGE SCAN | W_ORDER_D_N6 | 1 | 41 | 3 (34)| -----------------------------------------------------------------------------------------------
Could you please tell how to overcome this problem?
Thank you very much.joshic wrote:
Hi Randolph,The optimizer is set to 'choose', which is the default in 9i. So I guess that CBO uses the "all_rows" mode, because his stats are present.
According to your advice, I tried to adjust optimizer_mode, first_rows_1, first_rows_10, first_rows_100 and all_rows at the session level, but nothing helped.
Another question: since you are obviously having a schema star why don't you use not three separate bitmap indexes on the foreign key columns in the fact table? In this way, you can use a transformation star rather than using a composite index.
You use many OLTP, small, such as transactions that modify your fact table, or is it only loaded by loading in bulk? If you use only loads in bulk index bitmap should be fine.
Kind regards
RandolfOracle related blog stuff:
http://Oracle-Randolf.blogspot.com/SQLTools ++ for Oracle (Open source Oracle GUI for Windows):
http://www.sqltools-plusplus.org:7676 /.
http://sourceforge.NET/projects/SQLT-pp/ -
Hello world
I have this recommendation in Enterprise Manager, I would like to implement the first recommendation where there is a best execution plan that comes with the benefit of 99.7%. Its easy to implement using Enterprise manager by clicking just. But I would like to know how I can change execution plan using sqlplus. And if I want to return to the original plan, how can I do? Thank you in advance for your precious time really appreciate it.
Yes, you can restore it-here, you will...
http://docs.Oracle.com/CD/B28359_01/server.111/b28274/sql_tune.htm#CHDIDBBG
-
Cost of the execution plan is different in the two databases
Hi gurus,
I have two databases, which are 12 C.
The execution plan cost is different in the two databases for the same query.
is it possible to copy the execution to another plan.
Thank you in advance
Kind regards
REDA
Jr.Raj wrote:
There are a few differences.
The machine that has a high cost, has setup up, linux, 2-node RAC, more CPU.
and
other machine windows server less Setup and less cpus.
Please explain, cost really makes a difference.
Thank you & best regards
REDA
You simply can't compare costs like that. It does not work like that. Especially through two different systems. Here is a very good read: column the COST of the PLAN to EXPLAIN. Oracle FAQ
-
"Execution plan is not available" in OEM.
Greetings. Try to watch my first Exec Plan here and make the message displayed above in OEM 11 g even executing simple statements like below. I googled this and see only one question without an answer identical to this.
Thank you!
explain plan for
Select * from SCOTT. DEPT;
Don't know why you're looking for a plan in OEM.
* Connect to the base of data with the help of sqlplus and wwho and extract the map below;
SQL > explain plan for select * from SCOTT. DEPT;
SQL > SELECT * FROM TABLE (dbms_xplan.display);
* Very easy, set autotrace on and run the query, you will get the plan and statistics;
SQL > set autotrace
Syntax: SET AUTOT [RACE] {OFF |} WE | TRACE [ONLY]} [EXP [LAIN]] [[C] STAT]
SQL > set autotrace on
SQL > select * from double;
D
-
X
Execution plan
----------------------------------------------------------
Hash value of plan: 272002086
--------------------------------------------------------------------------
| ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1. 2. 2 (0) | 00:00:01 |
| 1. TABLE ACCESS FULL | DOUBLE | 1. 2. 2 (0) | 00:00:01 |
--------------------------------------------------------------------------
Statistics
----------------------------------------------------------
1 recursive calls
0 db block Gets
Gets 2 compatible
0 physical reads
0 redo size
522 bytes sent via SQL * Net to client
523 bytes received via SQL * Net from client
2 SQL * Net back and forth to and from the client
0 sorts (memory)
0 sorts (disk)
1 rows processed
HTH,
Pradeep
Maybe you are looking for
-
Firefox still leaves us the div overflow style scroll bars. Chrome and they are passing firefox upward! Do you think that one day before I got to the old code, they will add a webkit for styling the div overflow scroll bars? I see this question is ab
-
WNCE 2001 cant access admin homepage
Hello I changed my router and I wanted to change the settings on my wnce2001. I tried to access the admin page by connecting the device to my pc. On the user and the password page, I tried all the possibilities (admin, admin, admin, password, admin,
-
Re: Satellite A200 - 1 M 4: starting very long procedure
Well, when I turn on my laptop it took a few minutes (4-5) before even Vista begins to work. I have this screen where Toshiba check my system and it stay that the audit of about 5 minutes. I thought that's the problem with Windows, but it's not. Some
-
FFT on several channels in FPGA
I have a client with a 9076 (Spartan-6 LX45) who wishes to perform the FFT on several channels of a module of 9205. I never run FFT on several channels and never run out of space on the FPGA before, but I strongly suspect that this could change that.
-
C00D11B1 error after installation of SP2 code.
Recently, I added internet access at home. I have updated the files of the window recently, including the addition of the Service Pack. I also did a disk cleanup. I tried to use Pandora (well before the update) and I also checked another CD I have