search
HomeDatabaseMysql TutorialOracle 11g R2 全表扫描成本计算(非工作量模式-noworkload)

数据库版本Oracle11gR2SQLgt; select * from v$version where rownum=1;BANNEROracle Database 11g Enterprise Edition Release

数据库版本Oracle11gR2
SQL> select * from v$version where rownum=1;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production

创建手动管理的表空间,blockssize 8k
SQL> create tablespace test datafile
'/u01/app/oracle/oradata/ROBINSON/datafile/test.dbf' size 50m autoextend on maxsize 200m
uniform size 1m segment space management manual blocksize 8k;  2    3

Tablespace created.

创建测试用户test,默认表空间 test
SQL> create user test identified by oracle default tablespace test;

User created.

为了简便,授权DBA给test
SQL> grant dba to test;

Grant succeeded.

创建测试表test
SQL> create table test as select * from dba_objects where 1=0 ;

Table created.

设置pctfree 99
SQL> alter table test pctfree 99 pctused 1;

Table altered.

SQL> insert into test select * from dba_objects where rownum

1 row created.

确保一行一个block
SQL> alter table test minimize records_per_block;

Table altered.

SQL> insert into test select * from dba_objects where rownum

999 rows created.

SQL> commit;

Commit complete.

收集表统计信息
SQL> BEGIN
DBMS_STATS.GATHER_TABLE_STATS(ownname => 'TEST',
tabname => 'TEST',
estimate_percent => 100,
method_opt => 'for all columns size 1',
degree => DBMS_STATS.AUTO_DEGREE,
cascade=>TRUE
);
END;
/  2    3    4    5    6    7    8    9   10

PL/SQL procedure successfully completed.

SQL> select owner,blocks from dba_tables where owner='TEST' and table_name='TEST';

OWNER                              BLOCKS
------------------------------ ----------
TEST                                 1000

SQL> show parameter db_file_multiblock_read_count

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
db_file_multiblock_read_count        integer     16

全表扫描的成本等于220
SQL> select count(*) from test;

Execution Plan
----------------------------------------------------------
Plan hash value: 1950795681

-------------------------------------------------------------------
| Id  | Operation          | Name | Rows  | Cost (%CPU)| Time     |
-------------------------------------------------------------------
|   0 | SELECT STATEMENT   |      |     1 |   220   (0)| 00:00:03 |
|   1 |  SORT AGGREGATE    |      |     1 |            |          |
|   2 |   TABLE ACCESS FULL| TEST |  1000 |   220   (0)| 00:00:03 |
-------------------------------------------------------------------

成本的计算方式如下:
Cost = (
       #SRds * sreadtim +
       #MRds * mreadtim +
       CPUCycles / cpuspeed
       ) / sreadtime
      
#SRds - number of single block reads
#MRds - number of multi block reads
#CPUCyles - number of CPU cycles

sreadtim - single block read time
mreadtim - multi block read time
cpuspeed - CPU cycles per second

注意:如果没有收集过系统统计信息,那么Oracle采用非工作量统计, 如果收集了,,Oracle采用工作量统计的计算方法
SQL> select pname, pval1 from sys.aux_stats$ where sname='SYSSTATS_MAIN';

PNAME                               PVAL1
------------------------------ ----------
CPUSPEED
CPUSPEEDNW                     2696.05568
IOSEEKTIM                              10
IOTFRSPEED                           4096
MAXTHR
MBRC
MREADTIM
SLAVETHR
SREADTIM

9 rows selected.

我这里因为MBRC 为0,所以CBO采用了非工作量(noworkload)来计算成本

#SRds=0,因为是全表扫描,单块读为0
#MRds=表的块数/多块读参数=1000/16

mreadtim=ioseektim+db_file_multiblock_count*db_block_size/iotftspeed
SQL> select (select pval1 from sys.aux_stats$ where pname = 'IOSEEKTIM') +
  2         (select value
          from v$parameter
         where name = 'db_file_multiblock_read_count') *
       (select value from v$parameter where name = 'db_block_size') /
       (select pval1 from sys.aux_stats$ where pname = 'IOTFRSPEED') "mreadtim"
  3    4    5    6    7    from dual;

  mreadtim
----------
        42

sreadtim=ioseektim+db_block_size/iotfrspeed
SQL> select (select pval1 from sys.aux_stats$ where pname = 'IOSEEKTIM') +
       (select value from v$parameter where name = 'db_block_size') /
       (select pval1 from sys.aux_stats$ where pname = 'IOTFRSPEED') "sreadtim"
  from dual;  2    3    4

  sreadtim
----------
        12       
       
CPUCycles 等于 PLAN_TABLE里面的CPU_COST

SQL> explain plan for select count(*) from test;

Explained.

SQL> select cpu_cost from plan_table;

  CPU_COST
----------
   7271440

cpuspeed 等于 CPUSPEEDNW= 2696.05568

那么COST=1000/16*42/12+7271440/2696.05568/12/1000

SQL>  select ceil(1000/16*42/12+7271440/2696.05568/12/1000) from dual;

CEIL(1000/16*42/12+7271440/2696.05568/12/1000)
----------------------------------------------
                                           219

手工计算出来的COST用四舍五入等于219,和我们看到的220有差别, 这是由于隐含参数_tablescan_cost_plus_one参数造成的

SQL> SELECT x.ksppinm NAME, y.ksppstvl VALUE, x.ksppdesc describ
 FROM x$ksppi x, x$ksppcv y
  WHERE x.inst_id = USERENV ('Instance')
   AND y.inst_id = USERENV ('Instance')
   AND x.indx = y.indx
   AND x.ksppinm LIKE '%_table_scan_cost_plus_one%'
/  2    3    4    5    6    7

NAME                           VALUE      DESCRIB
------------------------------ ---------- ------------------------------
_table_scan_cost_plus_one      TRUE       bump estimated full table scan
                                           and index ffs cost by one
根据该参数的描述,在table full scan和index fast full scan的时候会将cost+1
那么我把改参数禁止了试一试

SQL> alter session set "_table_scan_cost_plus_one"=false;

Session altered.

SQL> set autot trace
SQL> select count(*) from test;

Execution Plan
----------------------------------------------------------
Plan hash value: 1950795681

-------------------------------------------------------------------
| Id  | Operation          | Name | Rows  | Cost (%CPU)| Time     |
-------------------------------------------------------------------
|   0 | SELECT STATEMENT   |      |     1 |   219   (0)| 00:00:03 |
|   1 |  SORT AGGREGATE    |      |     1 |            |          |
|   2 |   TABLE ACCESS FULL| TEST |  1000 |   219   (0)| 00:00:03 |
-------------------------------------------------------------------

这次得到的Cost等于219,与计算值正好匹配,现在更改db_file_multiblock_read_count参数

SQL> alter session set db_file_multiblock_read_count=32;

Session altered.

这个时候 sreadtim=12

SQL> select (select pval1 from sys.aux_stats$ where pname = 'IOSEEKTIM') +
       (select value from v$parameter where name = 'db_block_size') /
       (select pval1 from sys.aux_stats$ where pname = 'IOTFRSPEED') "sreadtim"
  from dual;  2    3    4

  sreadtim
----------
        12

mreadtim=74      
       
SQL> select (select pval1 from sys.aux_stats$ where pname = 'IOSEEKTIM') +
       (select value
  2    3            from v$parameter
  4           where name = 'db_file_multiblock_read_count') *
  5         (select value from v$parameter where name = 'db_block_size') /
  6         (select pval1 from sys.aux_stats$ where pname = 'IOTFRSPEED') "mreadtim"
  7    from dual;

  mreadtim
----------
        74

那么cost等于

SQL> select ceil(1000/32*74/12+7271440/2696.05568/12/1000) from dual;

CEIL(1000/32*74/12+7271440/2696.05568/12/1000)
----------------------------------------------
                                           193
SQL> set autot trace
SQL> select count(*) from test;

Execution Plan
----------------------------------------------------------
Plan hash value: 1950795681

-------------------------------------------------------------------
| Id  | Operation          | Name | Rows  | Cost (%CPU)| Time     |
-------------------------------------------------------------------
|   0 | SELECT STATEMENT   |      |     1 |   193   (0)| 00:00:03 |
|   1 |  SORT AGGREGATE    |      |     1 |            |          |
|   2 |   TABLE ACCESS FULL| TEST |  1000 |   193   (0)| 00:00:03 |
-------------------------------------------------------------------

与计算的Cost相匹配,从实验种可以得出,在11gR2中,全表扫描计算Cost的方式依然和9i/10g一样,没有变化。

相关链接

linux

Statement
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
MySQL: BLOB and other no-sql storage, what are the differences?MySQL: BLOB and other no-sql storage, what are the differences?May 13, 2025 am 12:14 AM

MySQL'sBLOBissuitableforstoringbinarydatawithinarelationaldatabase,whileNoSQLoptionslikeMongoDB,Redis,andCassandraofferflexible,scalablesolutionsforunstructureddata.BLOBissimplerbutcanslowdownperformancewithlargedata;NoSQLprovidesbetterscalabilityand

MySQL Add User: Syntax, Options, and Security Best PracticesMySQL Add User: Syntax, Options, and Security Best PracticesMay 13, 2025 am 12:12 AM

ToaddauserinMySQL,use:CREATEUSER'username'@'host'IDENTIFIEDBY'password';Here'showtodoitsecurely:1)Choosethehostcarefullytocontrolaccess.2)SetresourcelimitswithoptionslikeMAX_QUERIES_PER_HOUR.3)Usestrong,uniquepasswords.4)EnforceSSL/TLSconnectionswith

MySQL: How to avoid String Data Types common mistakes?MySQL: How to avoid String Data Types common mistakes?May 13, 2025 am 12:09 AM

ToavoidcommonmistakeswithstringdatatypesinMySQL,understandstringtypenuances,choosetherighttype,andmanageencodingandcollationsettingseffectively.1)UseCHARforfixed-lengthstrings,VARCHARforvariable-length,andTEXT/BLOBforlargerdata.2)Setcorrectcharacters

MySQL: String Data Types and ENUMs?MySQL: String Data Types and ENUMs?May 13, 2025 am 12:05 AM

MySQloffersechar, Varchar, text, Anddenumforstringdata.usecharforfixed-Lengthstrings, VarcharerForvariable-Length, text forlarger text, AndenumforenforcingdataAntegritywithaetofvalues.

MySQL BLOB: how to optimize BLOBs requestsMySQL BLOB: how to optimize BLOBs requestsMay 13, 2025 am 12:03 AM

Optimizing MySQLBLOB requests can be done through the following strategies: 1. Reduce the frequency of BLOB query, use independent requests or delay loading; 2. Select the appropriate BLOB type (such as TINYBLOB); 3. Separate the BLOB data into separate tables; 4. Compress the BLOB data at the application layer; 5. Index the BLOB metadata. These methods can effectively improve performance by combining monitoring, caching and data sharding in actual applications.

Adding Users to MySQL: The Complete TutorialAdding Users to MySQL: The Complete TutorialMay 12, 2025 am 12:14 AM

Mastering the method of adding MySQL users is crucial for database administrators and developers because it ensures the security and access control of the database. 1) Create a new user using the CREATEUSER command, 2) Assign permissions through the GRANT command, 3) Use FLUSHPRIVILEGES to ensure permissions take effect, 4) Regularly audit and clean user accounts to maintain performance and security.

Mastering MySQL String Data Types: VARCHAR vs. TEXT vs. CHARMastering MySQL String Data Types: VARCHAR vs. TEXT vs. CHARMay 12, 2025 am 12:12 AM

ChooseCHARforfixed-lengthdata,VARCHARforvariable-lengthdata,andTEXTforlargetextfields.1)CHARisefficientforconsistent-lengthdatalikecodes.2)VARCHARsuitsvariable-lengthdatalikenames,balancingflexibilityandperformance.3)TEXTisidealforlargetextslikeartic

MySQL: String Data Types and Indexing: Best PracticesMySQL: String Data Types and Indexing: Best PracticesMay 12, 2025 am 12:11 AM

Best practices for handling string data types and indexes in MySQL include: 1) Selecting the appropriate string type, such as CHAR for fixed length, VARCHAR for variable length, and TEXT for large text; 2) Be cautious in indexing, avoid over-indexing, and create indexes for common queries; 3) Use prefix indexes and full-text indexes to optimize long string searches; 4) Regularly monitor and optimize indexes to keep indexes small and efficient. Through these methods, we can balance read and write performance and improve database efficiency.

See all articles

Hot AI Tools

Undresser.AI Undress

Undresser.AI Undress

AI-powered app for creating realistic nude photos

AI Clothes Remover

AI Clothes Remover

Online AI tool for removing clothes from photos.

Undress AI Tool

Undress AI Tool

Undress images for free

Clothoff.io

Clothoff.io

AI clothes remover

Video Face Swap

Video Face Swap

Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

VSCode Windows 64-bit Download

VSCode Windows 64-bit Download

A free and powerful IDE editor launched by Microsoft

WebStorm Mac version

WebStorm Mac version

Useful JavaScript development tools

mPDF

mPDF

mPDF is a PHP library that can generate PDF files from UTF-8 encoded HTML. The original author, Ian Back, wrote mPDF to output PDF files "on the fly" from his website and handle different languages. It is slower than original scripts like HTML2FPDF and produces larger files when using Unicode fonts, but supports CSS styles etc. and has a lot of enhancements. Supports almost all languages, including RTL (Arabic and Hebrew) and CJK (Chinese, Japanese and Korean). Supports nested block-level elements (such as P, DIV),

SAP NetWeaver Server Adapter for Eclipse

SAP NetWeaver Server Adapter for Eclipse

Integrate Eclipse with SAP NetWeaver application server.

Notepad++7.3.1

Notepad++7.3.1

Easy-to-use and free code editor