标签：pass oci ble most occurs wrapper mini dynamic partition
We have gotten a lot of questions recently regarding how to gather and maintain optimizer statistics on large partitioned tables. The majority of these questions can be summarized into two topics:
This article will describe both of these issues and explain how you can address them both.
This is big topic so I recommend that you also check out the three-part series of posts on maintaining incremental statistics in partitioned tables.
Out of Range
Large tables are often decomposed into smaller pieces called partitions in order to improve query performance and ease of data management. The Oracle query optimizer relies on both the statistics of the entire table (global statistics) and the statistics of the individual partitions (partition statistics) to select a good execution plan for a SQL statement. If the query needs to access only a single partition, the optimizer uses only the statistics of the accessed partition. If the query access more than one partition, it uses a combination of global and partition statistics.
"Out of Range" means that the value supplied in a where clause predicate is outside the domain of values represented by the [minimum, maximum] column statistics. The optimizer prorates the selectivity based on the distance between the predicate value and the maximum value (assuming the value is higher than the max), that is, the farther the value is from the maximum value, the lower the selectivity will be. This situation occurs most frequently in tables that are range partitioned by a date column, a new partition is added, and then queried while rows are still being loaded in the new partition. The partition statistics will be stale very quickly due to the continuous trickle feed load even if the statistics get refreshed periodically. The maximum value known to the optimizer is not correct leading to the "Out of Range" condition. The under-estimation of selectivity often leads the query optimizer to pick a sub optimal plan. For example, the query optimizer would pick an index access path while a full scan is a better choice.
The "Out of Range" condition can be prevented by using the new copy table statistics procedure available in Oracle Database10.2.0.4 and 11g. This procedure copies the statistics of the source [sub] partition to the destination [sub] partition. It also copies the statistics of the dependent objects: columns, local (partitioned) indexes etc. It adjusts the minimum and maximum values of the partitioning column as follows; it uses the high bound partitioning value as the maximum value of the first partitioning column (it is possible to have concatenated partition columns) and high bound partitioning value of the previous partition as the minimum value of the first partitioning column for range partitioned table. It can optionally scale some of the other statistics like the number of blocks, number of rows etc. of the destination partition.
Assume we have a table called SALES that is ranged partitioned by quarter on the SALES_DATE column. At the end of every day data is loaded into latest partition. However, statistics are only gathered at the end of every quarter when the partition is fully loaded. Assuming global and partition level statistics (for all fully loaded partitions) are up to date, use the following steps in order to prevent getting a sub-optimal plan due to "out of range".
1. Lock the table statistics using LOCK_TABLE_STATS procedure in DBMS_STATS. This is to avoid interference from auto statistics job.EXEC DBMS_STATS.LOCK_TABLE_STATS(‘SH‘,‘SALES‘);2. Before beginning the initial load into each new partition (say SALES_Q4_2000) copy the statistics from the previous partition (say SALES_Q3_2000) using COPY_TABLE_STATS. You need to specify FORCE=>TRUE to override the statistics lock.
EXEC DBMS_STATS.COPY_TABLE_STATS (‘SH‘, ‘SALES‘, ‘SALES_Q3_2000‘, ‘SALES_Q4_2000‘, FORCE=>TRUE);
Expensive global statistics collection
In data warehouse environment it is very common to do a bulk load directly into one or more empty partitions. This will make the partition statistics stale and may also make the global statistics stale. Re-gathering statistics for the effected partitions and for the entire table can be very time consuming. Traditionally, statistics collection is done in a two-pass approach:
The full scan of the table for global statistics collection can be very expensive depending on the size of the table. Note that the scan of the entire table is done even if we change a small subset of partitions.
We avoid scanning the whole table when computing global statistics by deriving the global statistics from the partition statistics. Some of the statistics can be derived easily and accurately from partition statistics. For example, number of rows at global level is the sum of number of rows of partitions. Even global histogram can be derived from partition histograms. But the number of distinct values (NDV) of a column cannot be derived from partition level NDVs. So, Oracle maintains another structure called a synopsis for each column at the partition level. A synopsis can be considered as sample of distinct values. The NDV can be accurately derived from synopses. We can also merge multiple synopses into one. The global NDV is derived from the synopsis generated by merging all of the partition level synopses. To summarize:
Incremental maintenance feature is disabled by default. It can be enabled by changing the INCREMENTAL table preference to true. It can also be enabled for a particular schema or at the database level.
Assume we have table called SALES that is range partitioned by day on the SALES_DATE column. At the end of every day data is loaded into latest partition and partition statistics are gathered. Global statistics are only gathered at the end of every month because gathering them is very time and resource intensive. Use the following steps in order to maintain global statistics after every load.
Turn on incremental feature for the table.
EXEC DBMS_STATS.SET_TABLE_PREFS(‘SH‘,‘SALES‘,‘INCREMENTAL‘,‘TRUE‘);At the end of every load gather table statistics using GATHER_TABLE_STATS command. You don‘t need to specify the partition name. Also, do not specify the granularity parameter. The command will collect statistics for partitions where data has change or statistics are missing and update the global statistics based on the partition level statistics and synopsis.
EXEC DBMS_STATS.GATHER_TABLE_STATS(‘SH‘,‘SALES‘); （11g）
Note: that the incremental maintenance feature was introduced in Oracle Database 11g Release 1. However, we also provide a solution in Oracle Database10g Release 2 (10.2.0.4) that simulates the same behavior. The 10g solution is a new value, ‘APPROX_GLOBAL AND PARTITION‘ for the GRANULARITY parameter of the GATHER_TABLE_STATS procedures. It behaves the same as the incremental maintenance feature except that we don‘t update the NDV for non-partitioning columns and number of distinct keys of the index at the global level. For partitioned column we update the NDV as the sum of NDV at the partition levels. Also we set the NDV of columns of unique indexes as the number of rows of the table. In general, non-partitioning column NDV at the global level becomes stale less often. It may be possible to collect global statistics less frequently then the default (when table changes 10%) since approx_global option maintains most of the global statistics accurately.
分类： Linux操作系统
2014-10-13 11:17:09
In CBO, Statistic is critical important for Optimizer. Precise and timely statistics will truly reflect the data distribution and volume, and generate wiser SQL Execution Plan (SEP). After years of improvement, CBO is already widely accepted as the default and desirable solution for Oracle Optimizer.
As I mention in many blogs before, statistic is like the raw material for CBO optimizer. Without accurate statistic, it’s impossible to generate optimal SEP for CBO. So in CBO era, DBAs’ concern is about how to collect data statistic and finding out the proper frequency to collect them.
Since 10g, Oracle introduces a new feature of data statistic job to collect statistic automatically. The job will run at daily interval to collect statistic, and provide most database object statistic to CBO. The new feature solves many problems in most cases. The statistic collection job and mature dynamic sampling sweep the barrier of CBO’s usage.
Frankly speaking, Oracle internal functions are already enough for ordinary system requirement. But for some special cases, life would be tougher.
For many OLTP/OLAP systems, they usually load bulk of data into some extremely huge table (Mostly Partition Tables) in the night and some other business-free time, and then doing the processing work. The problem is that Oracle would have few chances to collect statistic after the data loading, which would influence the data volume and distribution information.
The problem would finally lead to the bad performance in the next processing work. In production environments, we often receive complains about some job processing is extremely slow. But after clearly examination, the old statistic is main reason. After collection, the job would be better. And then in the next loading, errors would happen again.
If we insert the statistic collection statement between data loading and processing, there would be some time consuming drawbacks for some big partition tables. For most partitions in these tables, the data is stable, and loading work only affects one or two partitions. But for normal partition statistic works, Oracle will collect all partition including inactive partitions, which consume a lot of resource.
In 11g, Oracle introduces a new feature named “Incremental Statistic”, which will only collect the partitions which are new or undergoing huge data changes. So it will save the time of collecting inactive partitions and make collection work shorter.
The new feature of “Incremental Statistic” is first introduced in the update edition of 10.2.0.4, and become formal in Oracle 11gR1. So we chose the Oracle 11gR2 as the test environment.
SQL> select * from v$version;
BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
PL/SQL Release 11.2.0.1.0 - Production
CORE 11.2.0.1.0 Production
I prepare a normal list partition table named T_PART.
SQL> create table t_part
2 ( owner varchar2(100),
3 object_id number,
4 object_name varchar2(100),
5 create_date date
6 )
7 partition by list(owner)
8 (partition t_part_pub values (‘PUBLIC‘),
9 partition t_part_sys values (‘SYS‘),
10 partition t_part_big1 values (‘APEX_030200‘,‘SYSMAN‘),
11 partition t_part_big2 values (‘ORDSYS‘,‘MDSYS‘),
12 partition t_part_other values(default));
Table created
Firstly, let’s see the default Oracle behavior. for partition statistic.
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘PUBLIC‘,‘ORDSYS‘);
30228 rows inserted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 27696 197 2012/10/9 22:08:11 YES
T_PART_SYS 0 0 2012/10/9 22:08:11 YES
T_PART_BIG1 0 0 2012/10/9 22:08:10 YES
T_PART_BIG2 2532 20 2012/10/9 22:08:10 YES
T_PART_OTHER 0 0 2012/10/9 22:08:10 YES
The default behavior. is that Oracle will collect all partitions according to the last_analyzed column in view dba_tab_partition.
Doing some modification focus on some partitions, let’s see the collection results.
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘SYS‘,‘APEX_030200‘);
33241 rows inserted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 27696 197 2012/10/9 22:13:31 YES
T_PART_SYS 30835 197 2012/10/9 22:13:31 YES
T_PART_BIG1 2406 20 2012/10/9 22:13:30 YES
T_PART_BIG2 2532 20 2012/10/9 22:13:31 YES
T_PART_OTHER 0 0 2012/10/9 22:13:31 YES
The insert statement only affects partition T_PART_SYS and T_PART_BIG1. But when we run the collection statement, all partitions will getting involved in the collection work.
It’s easy to change the statistic strategy for big partition table in Oracle 11g. Using the package dbms_stats will help setting the parameters.
The default strategy for Oracle partition table is not incremental setting. Only three parameters will affect the behavior.
SQL> select dbms_stats.get_prefs(‘PUBLISH‘,‘SYS‘,‘T_PART‘) from dual;
DBMS_STATS.GET_PREFS(‘PUBLISH‘
--------------------------------------------------------------------------------
TRUE
SQL> select dbms_stats.get_prefs(‘INCREMENTAL‘,‘SYS‘,‘T_PART‘) from dual;
DBMS_STATS.GET_PREFS(‘INCREMEN
--------------------------------------------------------------------------------
FALSE
SQL> select dbms_stats.get_prefs(‘GRANULARITY‘,‘SYS‘,‘T_PART‘) from dual;
DBMS_STATS.GET_PREFS(‘GRANULAR
--------------------------------------------------------------------------------
AUTO
Parameter “PUBLISH” is used to determinate whether CBO use the newest statistic when new statistic coming. It relates to pending statistic feature in most cases. And we need to ensure its value is true (Default Value).
Parameter “Incremental” default value is false, which means it would not adopt incremental statistic strategy by default.
Parameter “GRANULARITY” is not related to incremental statistic strategy in Oracle 11g, but related in Oracle 10gR2. So we need to keep it “AUTO” in Oracle 11g environment.
The estimated_precent should be keep to “AUTO”, in order to maintain the incremental statistic.
In the next section, we will see the effect of the new feature.
First, we need to change the setting for Big partition table.
SQL> exec dbms_stats.set_table_prefs(user,‘T_PART‘,pname => ‘INCREMENTAL‘,pvalue => ‘TRUE‘);
PL/SQL procedure successfully completed
SQL> select dbms_stats.get_prefs(‘INCREMENTAL‘,‘SYS‘,‘T_PART‘) from dual;
DBMS_STATS.GET_PREFS(‘INCREMEN
--------------------------------------------------------------------------------
TRUE
Do some modification work to T_PART, and see the effect of collection.
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 27696 197 2012/10/9 22:13:31 YES
T_PART_SYS 30835 197 2012/10/9 22:13:31 YES
T_PART_BIG1 2406 20 2012/10/9 22:13:30 YES
T_PART_BIG2 2532 20 2012/10/9 22:13:31 YES
T_PART_OTHER 0 0 2012/10/9 22:13:31 YES
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner not in (‘SYS‘,‘APEX_030200‘,‘PUBLIC‘,‘SYSMAN‘,‘ORDSYS‘,‘MDSYS‘);
3465 rows inserted
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘SYS‘);
30835 rows inserted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 27696 197 2012/10/9 22:29:26 YES
T_PART_SYS 61670 398 2012/10/9 22:29:21 YES
T_PART_BIG1 2406 20 2012/10/9 22:29:29 YES
T_PART_BIG2 2532 20 2012/10/9 22:29:22 YES
T_PART_OTHER 3465 23 2012/10/9 22:29:28 YES
The result is weird. Only two partitions are affected in the insert serial, but after collection, we find all partitions are collected. Not the things should be~
Let’s do more continue.
--Delete only One Partition Data
SQL> delete t_part where wner=‘SYS‘;
61670 rows deleted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 27696 197 2012/10/9 22:29:26 YES
T_PART_SYS 0 398 2012/10/9 22:42:47 YES
T_PART_BIG1 2406 20 2012/10/9 22:29:29 YES
T_PART_BIG2 2532 20 2012/10/9 22:29:22 YES
T_PART_OTHER 3465 23 2012/10/9 22:29:28 YES
The desired result comes out. Only the affected partitions is collected according to the last_analyzed column, and the table analyzed time is also changed.
SQL> select table_name, last_analyzed from dba_tables where wner=‘SYS‘ and table_name=‘T_PART‘;
TABLE_NAME LAST_ANALYZED
------------------------------ --------------------
T_PART 2012/10/9 22:42:49
If we affect more than one partition at once, how things would be.
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘PUBLIC‘);
27696 rows inserted
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘APEX_030200‘);
2406 rows inserted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 55392 388 2012/10/9 22:47:07 YES
T_PART_SYS 30874 398 2012/10/9 22:45:21 YES
T_PART_BIG1 4812 36 2012/10/9 22:47:08 YES
T_PART_BIG2 2532 20 2012/10/9 22:29:22 YES
T_PART_OTHER 3465 23 2012/10/9 22:29:28 YES
The truncate statement will affect the partitions and let’s see the results.
SQL> truncate table t_part;
Table truncated
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 0 0 2012/10/9 22:55:31 YES
T_PART_SYS 0 0 2012/10/9 22:55:31 YES
T_PART_BIG1 0 0 2012/10/9 22:55:31 YES
T_PART_BIG2 0 0 2012/10/9 22:55:31 YES
T_PART_OTHER 0 0 2012/10/9 22:55:31 YES
SQL> insert into t_part select owner, object_id, object_name, sysdate from dba_objects where owner in (‘SYS‘,‘APEX_030200‘);
33280 rows inserted
SQL> commit;
Commit complete
SQL> exec dbms_stats.gather_table_stats(user,‘T_PART‘,cascade => true);
PL/SQL procedure successfully completed
SQL> select partition_name, NUM_ROWS, BLOCKS, LAST_ANALYZED, global_stats from dba_tab_partitions where table_owner=‘SYS‘ and table_name=‘T_PART‘;
PARTITION_NAME NUM_ROWS BLOCKS LAST_ANALYZED GLOBAL_STATS
--------------- ---------- ---------- -------------------- ------------
T_PART_PUB 0 0 2012/10/9 22:55:31 YES
T_PART_SYS 30874 202 2012/10/9 22:58:05 YES
T_PART_BIG1 2406 20 2012/10/9 22:58:07 YES
T_PART_BIG2 0 0 2012/10/9 22:55:31 YES
T_PART_OTHER 0 0 2012/10/9 22:55:31 YES
Ok, the result is correct.
Only one thing we still wandering: Why we first switch the table to incremental statistic mode, and then we do some modification on some partitions. After collection, we found Oracle will collect the statistic for all partition, and then incremental will happen in the next collection.
The reasonable explanation is: Like Oracle incremental backup strategy, although you have a full mode backup, Oracle would do the incremental full backup work at first time. As for incremental statistic, things are similar. Oracle first needs a full incremental statistic, and then collects the increments.
The advantage of incremental statistic is that they can collect only active partitions’ data in the shorter time. The inactive partitions which do not have significant data modification operation will not getting involved in the collection work.
For the load data and then process jobs, it is the desirable collection strategy.
AUTO to gather subpartition, partition, or global statistics depending on the partition type.
###########3
深入分区表的增量统计信息收集技术(incremetal statistics collection)
http://blog.itpub.net/53956/viewspace-1372944
############5
Oracle Database - Enterprise Edition - Version 11.2.0.2 and later
Information in this document applies to any platform.
select substrb(dump(val,16,0,32),1,120) ep, cnt
from
(select /*+ no_expand_table(t) index_rs(t) no_parallel(t) no_parallel_index(t) dbms_stats cursor_sharing_exact use_weak_name_resl dynamic_sampling(0) no_monitoring no_substrb_pad */max(substrb("CLIENT_ID",1,32)) val,count(*) cnt
from "SYS"."WRH$_ACTIVE_SESSION_HISTORY" sample ( .3361644492) t
where substrb("CLIENT_ID",1,32) is not null ...
Depending on how much AWR data has been kept in the database, "SYS"."WRH$_ACTIVE_SESSION_HISTORY" table can be one of the largest SYS objects in some databases. As a result, it can take long time to gather statistics for this table.
1.Exclude any AWR repository table like WRH$_ACTIVE_SESSION_HISTORY which is known to take a long time to gather statistics:
2. Gather stats only for the objects with 10% or more data changes using options => ‘GATHER STALE option:
$ sqlplus / as sysdba
EXEC DBMS_STATS.GATHER_DICTIONARY_STATS ( options => ‘GATHER STALE‘);
标签：pass oci ble most occurs wrapper mini dynamic partition
原文地址：https://www.cnblogs.com/feiyun8616/p/9193034.html
