Using Adaptive Cursor Sharing (ACS) to produce

Report
1
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Using Adaptive Cursor
Sharing (ACS) to produce
multiple Optimal Plans
Carlos Sierra
Consulting Technical Advisor
Carlos Sierra
Oracle Server Technologies(ST) Center of Expertise(CoE)
 SQL Tuner handyman: developer, advisor, trainer, support
 IT: Oracle(17), UNISYS(12), Ford(3), others(3)
 Florida(17), Venezuela(3), Puerto Rico(6), Michigan(1), Mexico(X)
 Tools: SQLTXPLAIN(SQLT), SQLHC, TRCANLZR(TRCA), others
 Motto: Life is good!
3
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Adaptive Cursor Sharing (ACS)
 Motivation
 Mechanics
 Test Case
 Demo
 Remarks
4
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
ACS Motivation
5
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
SQL Processing
Hard parsing is expensive!
 Hard parse side effects
– CPU consumption
– Latch contention
 Excessive hard parsing
– Affects concurrency
– Restricts scalability
 Mitigating hard parsing
– Cursor sharing
6
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Implementing Cursor Sharing
Replacing literals with bind variables
7
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Before Bind Peeking
Before 9i CBO was blind to values passed
 Predicate
– WHERE channel_id = :b1
 Unknowns
– Is :b1 between low and high values of channel_id?
– Is :b1 a popular value of channel_id?
– Are there any rows with value :b1 for channel_id?
 Penalty
– Possible suboptimal plans
8
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
With Bind Peeking
9i offers a partial solution
 Predicate
– WHERE channel_id = :b1
 Plan is determined by peeked values
– EXEC :b1 := 9;
 Optimal plan for 1st execution
– CBO can use low/high and histograms on channel_id
 Penalty
– Possible suboptimal plans for subsequent executions on skewed data
9
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
With Adaptive Cursor Sharing
11g improves cursor sharing
 Some queries are ACS candidates
 Sophisticated non-persistent mechanism
 Selectivity of predicates determine plan
 Multiple optimal plans for a query!
– If ACS is successfully applied
 Penalty
– Marginal increase in CPU and memory overhead
10
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
ACS Mechanics
11
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
ACS high-level Overview
High level overview
 If SQL with binds meets some requirements
– Flag cursor as bind sensitive
– Start monitoring data volume manipulated by cursor
 If bind sensitive and data volume manipulated by cursor varies
significantly
– Flag cursor as bind aware
– Start generating multiple optimal plans for this query on next hard parse
 If bind aware then use selectivity of predicates to decide on plan
12
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bind Sensitive
Minimum requirements
 SQL has explicit binds
– Or literals and cursor_sharing is “force”
 Predicate: column + operand + bind_variable
– Equality operand “=“ and histogram on column
 Ex: channel_id = :b1
– Non-equality operand (range) regardless of histogram on column
 “>”, “>=“, “<“, ‘<=“, BETWEEN, LIKE
13
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bind Aware
How to become bind aware?
 Significant changes in data volume manipulated by cursor
– A few rows versus a few thousands of rows
– A few thousands of rows versus a few millions of rows
 Specifying /*+ BIND_AWARE */ CBO Hint
– Bypasses the monitoring phase on data volume
14
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Plan Selection
Based on selectivity profile of predicates
 Evaluate selectivity of predicates at soft parse
 Compare to a non-persistent selectivity profile
 If within ranges of a known profile then select associated plan
 Else hard parse
– Compute and execute newly generated plan
– Create selectivity profile for new plan or update profile of existing plan
 If ranges on selectivity profiles overlap then merge profiles
15
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
V$ dynamic views for ACS
ACS non-persistent performance views
 V$SQL
– Shareable, bind sensitive and bind aware flags
 V$SQL_CS_STATISTICS
– Data volume manipulated (rows processed)
 V$SQL_CS_HISTOGRAM
– Record keeping of data volume per execution (small, medium, large)
 V$SQL_CS_SELECTIVITY
– Predicates selectivity profiles
16
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
ACS Test Case
17
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Our Query with Literals
Guesstimate execution plan then verify it with demo 0
18
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Possible Access Paths?
19
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Optimal Execution Plan
20
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Our Query with Bind Variables
How many optimal execution plans can you foresee?
21
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Multiple Optimal Plans for one Query
Guesstimate optimal plan (access paths) for each query
22
Query
:b1
:b2
AP1
AP2
q1
9
33
N1
N2
q2
5
32
q3
2
999
q4
9
999
q5
2
33
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Multiple Optimal Plans for one Query
Guesstimate optimal plan (access paths) for each query
23
Query
:b1
:b2
AP1
AP2
q1
9
33
N1
N2
q2
5
32
N1
N2
q3
2
999
FTS
FTS
q4
9
999
N1
FTS
q5
2
33
FTS
N2
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Multiple Optimal Plans for one Query
Execute demos 1-5 and verify access paths
24
Query
:b1
:b2
AP1
AP2
q1
9
33
N1
N2
q2
5
32
N1
N2
q3
2
999
FTS
FTS
q4
9
999
N1
FTS
q5
2
33
FTS
N2
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bind Sensitive: Rows Processed
Monitor v$sql_cs_statistics.rows_processed
 Data volume manipulated
– Fuzzy representation
– S: few rows
– M: thousands or rows
– L: millions of rows
 v$sql_cs_histogram
– Bucket(0): S
– Bucket(1): M
– Bucket(2): L
25
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Rows
Processed
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
37,382
q2
5
32
N1/N2
2
q3
2
999 FTS/FTS
8,021,324
q4
9
999
N1/FTS
6,233,815
q5
2
33
FTS/N2
1,825,131
ACS Demo
26
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 6: When Cursor becomes Bind Aware?
Obtain rows processed from demo 1-5 then guesstimate aware flag
Rows
Processed
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
37,382
q2
5
32
N1/N2
2
q3
2
999 FTS/FTS
8,021,324
q4
9
999
N1/FTS
6,233,815
q5
2
33
FTS/N2
1,825,131
27
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bucket
Aware
Child
Actual
Demo 6: When Cursor becomes Bind Aware?
Obtain rows processed from demo 1-5 then guesstimate aware flag
Rows
Processed
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
37,382
1
q2
5
32
N1/N2
2
0
q3
2
999 FTS/FTS
8,021,324
2
q4
9
999
N1/FTS
6,233,815
2
q5
2
33
FTS/N2
1,825,131
2
28
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bucket
Aware
Child
Actual
Demo 6: When Cursor becomes Bind Aware?
Obtain rows processed from demo 1-5 then guesstimate aware flag
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
q2
5
32
N1/N2
q3
2
q4
q5
29
Rows
Processed
Bucket
Aware
37,382
1
N
2
0
N
999 FTS/FTS
8,021,324
2
Y
9
999
N1/FTS
6,233,815
2
Y
2
33
FTS/N2
1,825,131
2
Y
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Child
Actual
Demo 6: When Cursor becomes Bind Aware?
Obtain rows processed from demo 1-5 then guesstimate aware flag
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
q2
5
32
N1/N2
q3
2
q4
q5
30
Rows
Processed
Bucket
Aware
Child
Actual
37,382
1
N
0
N1/N2
2
0
N
0
N1/N2
999 FTS/FTS
8,021,324
2
Y
1
FTS/FTS
9
999
N1/FTS
6,233,815
2
Y
2
N1/FTS
2
33
FTS/N2
1,825,131
2
Y
3
FTS/N2
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 6: When Cursor becomes Bind Aware?
Obtain rows processed from demo 1-5 then guesstimate aware flag
Query
:b1
:b2
Optimal
q1
9
33
N1/N2
q2
5
32
N1/N2
q3
2
q4
q5
31
Rows
Processed
Bucket
Aware
Child
Actual
37,382
1
N
0
N1/N2
2
0
N
0
N1/N2
999 FTS/FTS
8,021,324
2
Y
1
FTS/FTS
9
999
N1/FTS
6,233,815
2
Y
2
N1/FTS
2
33
FTS/N2
1,825,131
2
Y
3
FTS/N2
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 7: When Cursor becomes Bind Aware?
Compute bucket and guesstimate aware flag and actual plan
Rows
Processed
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
1,825,131
q4
9
999
N1/FTS
6,233,815
q3
2
999 FTS/FTS
8,021,324
q2
5
32
N1/N2
2
q1
9
33
N1/N2
37,382
32
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bucket
Aware
Child
Actual
Demo 7: When Cursor becomes Bind Aware?
Compute bucket and guesstimate aware flag and actual plan
Rows
Processed
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
1,825,131
2
q4
9
999
N1/FTS
6,233,815
2
q3
2
999 FTS/FTS
8,021,324
2
q2
5
32
N1/N2
2
0
q1
9
33
N1/N2
37,382
1
33
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bucket
Aware
Child
Actual
Demo 7: When Cursor becomes Bind Aware?
Compute bucket and guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
1,825,131
2
N
N1/FTS
6,233,815
2
N
2
999 FTS/FTS
8,021,324
2
N
q2
5
32
N1/N2
2
0
N
q1
9
33
N1/N2
37,382
1
Y
34
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Child
Actual
Demo 7: When Cursor becomes Bind Aware?
Compute bucket and guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
Child
Actual
1,825,131
2
N
0
FTS/N2
N1/FTS
6,233,815
2
N
0
FTS/N2
2
999 FTS/FTS
8,021,324
2
N
0
FTS/N2
q2
5
32
N1/N2
2
0
N
0
FTS/N2
q1
9
33
N1/N2
37,382
1
Y
1
N1/N2
35
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 7: When Cursor becomes Bind Aware?
Compute bucket and guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
Child
Actual
1,825,131
2
N
0
FTS/N2
N1/FTS
6,233,815
2
N
0
FTS/N2
2
999 FTS/FTS
8,021,324
2
N
0
FTS/N2
q2
5
32
N1/N2
2
0
N
0
FTS/N2
q1
9
33
N1/N2
37,382
1
Y
1
N1/N2
36
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 8: When Cursor becomes Bind Aware?
Guesstimate aware flag and actual plan
Rows
Processed
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
1,825,131
2
q4
9
999
N1/FTS
6,233,815
2
q3
2
999 FTS/FTS
8,021,324
2
q1
9
33
N1/N2
37,382
1
q2
5
32
N1/N2
2
0
37
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Bucket
Aware
Child
Actual
Demo 8: When Cursor becomes Bind Aware?
Guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
1,825,131
2
N
N1/FTS
6,233,815
2
N
2
999 FTS/FTS
8,021,324
2
N
q1
9
33
N1/N2
37,382
1
N
q2
5
32
N1/N2
2
0
N
38
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Child
Actual
Demo 8: When Cursor becomes Bind Aware?
Guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
Child
Actual
1,825,131
2
N
0
FTS/N2
N1/FTS
6,233,815
2
N
0
FTS/N2
2
999 FTS/FTS
8,021,324
2
N
0
FTS/N2
q1
9
33
N1/N2
37,382
1
N
0
FTS/N2
q2
5
32
N1/N2
2
0
N
0
FTS/N2
39
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Demo 8: When Cursor becomes Bind Aware?
Guesstimate aware flag and actual plan
Query
:b1
:b2
Optimal
q5
2
33
FTS/N2
q4
9
999
q3
Rows
Processed
Bucket
Aware
Child
Actual
1,825,131
2
N
0
FTS/N2
N1/FTS
6,233,815
2
N
0
FTS/N2
2
999 FTS/FTS
8,021,324
2
N
0
FTS/N2
q1
9
33
N1/N2
37,382
1
N
0
FTS/N2
q2
5
32
N1/N2
2
0
N
0
FTS/N2
40
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Closing Remarks
41
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Remarks on Bind Sensitivity
Based on experimental observation
 Monitor V$SQL_CS_STATISTICS.rows_processed
– If small number of rows then
 V$SQL_CS_HISTOGRAM.bucket_id(0)++
– If medium number of rows then
 V$SQL_CS_HISTOGRAM.bucket_id(1)++
– If large number of rows then
 V$SQL_CS_HISTOGRAM.bucket_id(2)++
42
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Remarks on Bind Aware
Based on experimental observation
 Some cases where cursor may become bind aware
– bucket_id(0) = bucket_id(1) > 0
– bucket_id(1) = bucket_id(2) > 0
– bucket_id(0) > 0 and bucket_id(2) > 0
– /*+ BIND_AWARE */ CBO Hint
43
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Understanding Selectivity Profile
From demo 6
44
Query
:b1
:b2
Child
q3
2
999
1
q4
9
999
2
q5
2
33
3
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Understanding Selectivity Profile
From demo 6
45
Query
:b1
:b2
Child
q3
2
999
1
q4
9
999
2
q5
2
33
3
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Understanding Selectivity Profile
From demo 6
46
Query
:b1
:b2
Child
q3
2
999
1
q4
9
999
2
q5
2
33
3
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Conclusions
ACS can produce multiple optimal plans for one query
 ACS only applies to a subset of queries with binds
 ACS requires a ramp-up process (few executions)
 In some cases cursor may fail to become bind aware
 To force a cursor become bind aware use CBO Hint
 ACS is not persistent
 ACS works well with SQL Plan Management
47
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
Give Away
Script sqlt/utl/coe_gen_sql_patch.sql (MOS 215187.1)
 Creates a SQL Patch for one SQL_ID
 Turns “on” EVENT 10053 for SQL_ID
 Hints on SQL Patch
– GATHER_PLAN_STATISTICS
– MONITOR
– BIND_AWARE
 Instructions how to drop SQL Patch and turn “off” EVENT 10053
48
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
References and Contact Info
Oracle Optimizer Blog
 https://blogs.oracle.com/optimizer/
– Insight into the workings of the Optimizer
 [email protected]
 http://carlos-sierra.net
 @csierra_usa
49
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.
50
Copyright © 2013, Oracle and/or its affiliates. All rights reserved.

similar documents