Lecture Series on Relational Database Technology

Transcription

1 WI Department of Information Systems Lecture Series on Relational Database Technology Introduction into key Concepts of Relational DataBase Management Systems (RDBMS) Summer Term 2011 Eberhard Hechler SCITA, Executive IT Architect IBM Boeblingen Research & Development Lab

2 Content Day 1 and Day 2 Day Lecture Topics Introduction & Lecture Objectives/Scope Overview of Relational DataBase Management Systems (RDBMS) Database Security Aspects (Athentication, Authorization, Priviledges,...) Database Objects (Data Types, Schemas, Tables, Views, Indexes, Buffer Pools,...) Database Objects (continuation) Database Administration (e.g. Data Storage Management,...) Basic SQL Statements (Data Retrieval, simple Queries, Examples,...) Page 2

3 WI Department of Information Systems Database Administration

4 This Chapter s Agenda Database Objects Buffer Pools and Table Spaces Schemas Data Types Tables, Identity Columns and Temporary Tables Views Indexes Constraints Page 4

5 WI Department of Information Systems Database Objects

6 Database Objects Instance 1 dbm configuration file Lock List Database 1 Catalog Logs Database 2 Catalog Logs DB config file DB config file Table1 Table2 Table3 Table2 Lock List Index1 View1 Index1 View1 BLOBs View2 Index2 View3 Table spaces Page 6

7 WI Department of Information Systems Buffer Pools and Table Spaces

8 Buffer Pools - Overview Default Table spaces SYSCATSPACE TEMPSPACE1 USERSPACE1 MYREGSPACE MYTEMPSPACE MYLONGSPACE Buffer pools IBMDEFAULTBP MYBUFF1 MYBUFF2 MYBUFF3 Extended Storage EXTENDED STORAGE Global Database Memory Page 8

9 Management of Buffer Pools (1) Memory for the buffer pool is allocated when the database is activated (implicitly or explicitly) Command to create buffer pool CREATE BUFFERPOOL bpname [ IMMEDIATE DEFERRED ] SIZE sz PAGESIZE pgsz IMMEDIATE The buffer pool will be created immediately. If there is not enough reserved space in the database shared memory to allocate the new buffer pool, a warning (SQLSTATE 01657) is returned, and the statement is executed DEFERRED DEFERRED The buffer pool will be created when the database is reactivated (all applications need to be disconnected from the database) Page 9

10 Management of Buffer Pools (2) SIZE AUTOMATIC Using AUTOMATIC, the db manager will adapt the BP size depending on the workload CREATE BUFFERPOOL bpname [ SIZE sz SIZE [ sz ] AUTOMATIC ] PAGESIZE pgsz If no SIZE option is specified the BP is created as self-tuning with 1000 pg initial size Information stored in SYSIBM.SYSBUFFERPOOLS Page 10

11 Management of Buffer Pools (3) Use the ALTER BUFFERPOOL command to increase the size of the buffer pool, memory is allocated as soon as the command is committed if the memory is available. If the memory is not available, the change occurs when all applications are disconnected and the database is reactivated. If the size of the buffer pool is decreased, memory is deallocated at commit time Alter size of a buffer pool ALTER BUFFERPOOL bpname [ IMMEDIATE DEFERRED ] SIZE sz Activate SIZE AUTOMATIC ALTER BUFFERPOOL bpname SIZE AUTOMATIC Page 11

12 Table Spaces - Overview Relate logical objects (tables, indexes) to physical storage (container) Tables and other objects are created in table spaces Table spaces comprised of containers Two types of space management System Managed Space (SMS) Database Managed Space (DMS) Page 12

13 Table Space Structural Units Page Size Smallest unit of I/O 4K (default), 8K, 16K and 32K pages One page size per table space Defined on table space creation cannot be changed Must match page size of associated buffer pool Extent 0 Extent 1 Page 0 Page 1 Page 2 Page 3 Extent Size Unit of space allocation 2 or more pages One extent size per table space Defined on table space creation cannot be changed Default specified by database configuration parameter DFT_EXTENT_SZ Should match underlying disk stripe unit size Pre-fetch Size 4Unit for pre-fetching 4Should be equal to or multiple of extent size 4Should match underlying disk stripe size 4Tuning parameter can be changed Page 13

14 Default Table Spaces Three table spaces always created when the database is created: SYSCATSPACE catalog table space, contains system catalog tables TEMPSPACE1 default system temporary table space USERSPACE1 default user data table space Default case Customized space management type and containers CREATE DATABASE sample CREATE DATABASE sample CATALOG TABLESPACE MANAGED BY SYSTEM USING ( 'c:\catdir1' ) USER TABLESPACE MANAGED BY DATABASE USING ( FILE 'c:\db2files\usertbsp1 100, FILE 'c:\db2files\usertbsp2 100 ) TEMP TABLESPACE MANAGED BY SYSTEM USING ( 'c:\tempspace' ) Page 14

15 WI Department of Information Systems Schemas

16 Schemas Overview Schema is database object Can be explicitly created and dropped Has privileges Schema name used as high order part of two-part (or fully qualified ) object names <schema name>.<simple name> E.g., table SMITH.TAB1 is different from JONES.TAB1 Default schema name CURRENT SCHEMA special register Contains current authorization ID per default Change with SET CURRENT SCHEMA or SET CURRENT SQLID command Page 16

17 Schema Creation Explicit creation Implicit creation Database privilege Reserved schema names CREATE SCHEMA myschema SYSIBM, SYSCAT, SYSSTAT, SYSFUN Schema names like SYSxxx cannot be created by user Schema SESSION should not be used, because of Declared Global Temporary Tables (DGTT) Page 17

18 SYS Schemas Created with every database and placed into the SYSCATSPACE table space SYSIBM Base catalogs Access not recommended SYSCAT SELECT authority GRANTed to PUBLIC Catalog Read-only Views Recommended way to obtain catalog information Page 18

19 SYS Schemas Created with every database and placed into the SYSCATSPACE table space SYSSTAT Updateable Catalog Views - Influence the Optimizer SYSFUN User-Defined Functions Page 19

20 WI Department of Information Systems Data Types

21 Data Types XML is new Datalink is no longer supported Page 21

22 LARGE Objects To store large character strings or files To store large binary strings or files Maximum size is 2 GB (1 GB for DBCLOBs) NOT LOGGED for objects > 1 GB direct read/write access, no bufferpool Action News DB2 By The Book Binary Large Object BLOB Character Large Object CLOB Double Byte Character Large Object DB2 DBCLOB see also DB2IC: SQL and XQuery limits Page 22

23 User-Defined Distinct Types CREATE DISTINCT TYPE pound AS INTEGER WITH COMPARISONS ; CREATE DISTINCT TYPE kilogram AS INTEGER WITH COMPARISONS ; CREATE TABLE person ( f_name VARCHAR (30), weight_p pound NOT NULL, weight_k kilogram NOT NULL ) ; SELECT f_name FROM person WHERE weight_p > pound(30); WHERE integer(weight_p) > 30; succeeds SELECT f_name FROM person WHERE weight_p > weight_k; succeeds Page 23 fails

24 WI Department of Information Systems Tables, Identity Columns and Temporary Tables

25 CREATE TABLE Command Connect to database first You must have SYSADM or DBADM authority or CREATETAB privilege on the database Example: connect to eddb; create table artists ( artno SMALLINT NOT NULL, name VARCHAR(50) WITH DEFAULT 'abc, classification CHAR(1) NOT NULL, bio CLOB(100K) LOGGED, picture BLOB(2M) NOT LOGGED COMPAT ) INDEX IN indtbsp LONG IN longtbsp IN datatbsp; Page 25

26 Where is table placed by default If a table is created without the IN clause, the table data (and its indexes and LOB data) will be placed: In the IBMDEFAULTGROUP table space (if it exists and if the page size is sufficient) In a user created table space which is of the smallest page size that is sufficient for the table Then it will go in USERSPACE1 (if it exists and has a sufficient page size) The IN, INDEX IN, and LONG IN clauses specify where table spaces regular data, index, and large objects are to be stored in Page 26

27 CREATE TABLE... LIKE Table columns have exact same names and attributes May specify table, view, or nickname One for one copy of columns no constraints, triggers, foreign keys or indexes copied data not copied Example: CREATE TABLE tab1new LIKE tab1 Page 27

28 Definition Only Table Query used to define table Can be subset of single table or combination of tables Table not populated Column attributes of defined table based upon referenced table Example: CREATE TABLE t1new AS (SELECT c1, c8, c10 FROM t1) DEFINITION ONLY Page 28

29 NULL Values A NULL value represents an unknown state The CREATE TABLE statement can contain the phrase NOT NULL following the definition of each column This will ensure that the column contains a known data value Can specify a default value if NULL is entered Page 29

30 NULL Values Example: CREATE TABLE staff ( id SMALLINT NOT NULL WITH DEFAULT 10, name VARCHAR(9), dept SMALLINT NOT NULL WITH DEFAULT 20, job CHAR(5), years SMALLINT, salary DECIMAL(7, 2), comm DECIMAL(7, 2) WITH DEFAULT ); system default value user-defined default value Page 30

31 System Default Values If a specific default value is not specified following the DEFAULT keyword, the system default value of the column data type is used For example: Numeric: 0 CHAR(x): x blanks VARCHAR(x): a string of length 0 BLOB: a string of length 0 Check the DB2 Command Reference under the 'ALTER TABLE' command for a complete list of data types' system default values NULL and 0-Length Data Value Compression System Default Value Compression If VALUE COMPRESSION is used, use the optional COMPRESS SYSTEM DEFAULT option to further reduce disk space usage Page 31

32 Some Useful Commands LIST TABLES List tables for the current user LIST TABLES FOR ALL List all tables defined in the database LIST TABLES FOR SCHEMA <schema> List tables for the specified schema DESCRIBE TABLE <table name> Show the structure of the specified table Example: DESCRIBE TABLE department Column Type Type name schema name Length Scale Nulls DEPTNO SYSIBM CHARACTER 3 0 No DEPTNAME SYSIBM VARCHAR 29 0 No MGRNO SYSIBM CHARACTER 6 0 Yes ADMRDEPT SYSIBM CHARACTER 3 0 No LOCATION SYSIBM CHARACTER 16 0 Yes Page 32

33 Restrict Drop Table ALTER TABLE tab1 ADD RESTRICT ON DROP DROP TABLE tab1 SQL0672N Operation DROP not allowed on table USER.TAB1 ALTER TABLE tab1 DROP RESTRICT ON DROP Page 33

34 Identity Columns A numeric column in a table which automatically generates a unique numeric value for each row that is inserted One Identity column per table maximum Values can be generated by DB2 always or by default GENERATED ALWAYS Values are always generated by DB2 Applications are not allowed to provide an explicit value GENERATED BY DEFAULT Values can be explicitly provided by an application or if no value is given, then DB2 generates one DB2 cannot guarantee uniqueness for the inserted column value Intended for data propagation, unload/reload of a table Page 34

35 Identity Columns Generated By Default Example CREATE TABLE inventory (partno INTEGER PRIMARY KEY GENERATED BY DEFAULT AS IDENTITY (START WITH 100 INCREMENT BY 1), description CHAR(20) ); COMMIT; INSERT INTO inventory VALUES (DEFAULT,'door'); INSERT INTO inventory (description) VALUES ('hinge'); INSERT INTO inventory VALUES (200,'window'); INSERT INTO inventory VALUES (102,'handle'); INSERT INTO inventory VALUES (101,'bolt'); COMMIT; --->inserts 100,door --->inserts 101,hinge --->inserts 200,window --->inserts 102,handle --->error, duplicate INSERT INTO inventory (description) VALUES ('lock'); INSERT INTO inventory (description) VALUES ('lock'); ROLLBACK; --->error, duplicate --->inserts 103,lock INSERT INTO inventory (description) VALUES ('frame'); COMMIT; --->inserts 104,frame Page 35

36 GENERATED BY DEFAULT Example select * from inventory CREATE TABLE inventory ( partno INTEGER PRIMARY KEY GENERATED BY DEFAULT AS IDENTITY (START WITH 100, INCREMENT BY 1), description CHAR(20) ); PARTNO DESCRIPTION door 101 hinge 102 window 104 frame INSERT INTO inventory VALUES (DEFAULT,'door'); INSERT INTO inventory (description) VALUES ('hinge'); INSERT INTO inventory VALUES (101,'bolt'); INSERT INTO inventory VALUES (102,'window'); COMMIT; ---> inserts 100,door ---> inserts 101,hinge ---> ERROR, duplicate partno ---> inserts 102,window INSERT INTO inventory (description) VALUES ('lock'); INSERT INTO inventory (description) VALUES ('lock'); ROLLBACK; ---> ERROR, duplicate partno ---> inserts 103,lock INSERT INTO inventory (description) VALUES ('frame'); COMMIT; ---> inserts 104,frame Page 36

37 NOT LOGGED INITIALLY Tables (1) Useful for situations needing to insert large amounts of data from alternate source (another table or file) Data inserted without logging Use when recovery of table not required Flush to log when mincommit reached/ commit successful 'Not Logged Initially ' not enabled Log Buffer New Row Old Row log With Logging Bufferpool New Row table Insert Dirty pages written by cleaners to disk when softmax or chngpgs_thresh reached 'Not Logged Initially ' is enabled Bufferpool New Row table No Logging Flush to disk when commit successful Page 37

38 NOT LOGGED INITIALLY Tables (2) To use this option, the table must first be created using.. CREATE TABLE table-name... NOT LOGGED INITIALLY To improve concurrency after the CREATE COMMIT should be issued ( 'Not Logged Initially' state turned off) To reactivate the 'not logged initially' state: ALTER TABLE table-name ACTIVATE NOT LOGGED INITIALLY; INSERT INTO... SELECT FROM... ; COMMIT; ( 'Not Logged Initially' state turned off) OR ALTER TABLE table-name ACTIVATE NOT LOGGED INITIALLY WITH EMPTY TABLE ; INSERT INTO... SELECT FROM... ; COMMIT; ( 'Not Logged Initially' state turned off) Page 38

39 Declared Global Temporary Tables (DGTT) (1/3) Created and used by an application and dropped (automatically) when the connection is reset Can only be accessed by the application that created the table No entry exists in any catalog table to avoid catalog contention If multiple applications create a table of the same name, each application has a unique instance of that declared temporary table No authority checking No table locking or row locking Page 39

40 Declared Global Temporary Tables (DGTT) (2/3) Minimal undo logging Support the rollback of data changes made to global temporary table NOT LOGGED option Hold intermediate results during complex processing Automatic cleanup Index support Any standard index can be created on a temporary table Statistics support RUNSTATS supported against the table Page 40

41 Declared Global Temporary Tables (DGTT) (3/3) Declared temporary tables reside in a user temporary table space which has to be defined prior to creating any declared temporary tables Schema SESSION is always used Page 41

42 Declared Global Temporary Tables Example CREATE USER TEMPORARY TABLESPACE apptemps MANAGED BY SYSTEM USING ('apptemps'); DECLARE GLOBAL TEMPORARY TABLE temployess LIKE employee NOT LOGGED; DECLARE GLOBAL TEMPORARY TABLE tempdept ( deptid CHAR(6), deptname CHAR(20) ) ON COMMIT DELETE ROWS NOT LOGGED ; DECLARE GLOBAL TEMPORARY TABLE tempprojects AS ( fullselect ) DEFINITION ONLY ON COMMIT PRESERVE ROWS NOT LOGGED WITH REPLACE IN TABLESPACE apptemps; Page 42

43 WI Department of Information Systems Views

44 Creating Views Nested view supported At least SELECT privilege on base tables of view required View information kept in: SYSCAT.VIEWS, SYSCAT.VIEWDEP, SYSCAT.TABLES Page 44

45 Creating Views CONNECT TO TESTDB; CREATE VIEW DEPTSALARY AS SELECT DEPTNO, DEPTNAME, SUM(SALARY) AS TOTSAL FROM PAYROLL GROUP BY DEPTNO,DEPTNAME; CREATE VIEW EMPSALARY AS SELECT EMPNO, EMPNAME, SALARY FROM PAYROLL, PERSONNEL WHERE EMPNO=EMPNUMB AND SALARY > SELECT * FROM DEPTSALARY DEPTNO DEPTNAME TOTSAL MANUFACTURING ADMINISTRATION MARKETING Page 45

46 Views With Check Option Specifies the constraint that every row that is inserted or updated through the view must conform to the definition of the view A row that does not conform to the definition of the view is a row that does not satisfy the search conditions of the view Example: CREATE VIEW emp_view2 ( empno, empname, deptno ) AS ( SELECT id, name FROM employee WHERE dept = 10 ) WITH CHECK OPTION; When this view is used to insert or update with new values, the WITH CHECK OPTION will restrict the input values for the dept column Page 46

47 CASCADED and LOCAL Check Option If a view is defined based on another view or a table with check constraints, it is possible to inherit or not to inherit the search condition, two options available: WITH CASCADED CHECK OPTION (default) WITH LOCAL CHECK OPTION Example: CREATE VIEW emp_view3 AS ( SELECT empno, empname, deptno FROM emp_view2 WHERE empno > 20 ) WITH CASCADED CHECK OPTION; Conditions deptno = 10 (defined with emp_view2) AND empno > 20 will be checked for insert and update operations against this view Page 47

48 CASCADED and LOCAL Check Option Example: CREATE VIEW emp_view4 AS ( SELECT empno, empname, deptno FROM emp_view3 WHERE name = 'Smith' ) WITH LOCAL CHECK OPTION; Only condition name='smith' (defined in emp_view4) is checked for inserts and updates Page 48

49 WI Department of Information Systems Indexes

50 Creating Indexes Index Characteristics: Ascending or descending Unique or non-unique Compound Cluster Bi-directional - may specify ALLOW or DISALLOW REVERSE SCANS Include columns - may only be used if UNIQUE is specified Local range partitioned indexes New in DB2 v9.7 Page 50

51 Creating Indexes Examples: create unique index itemno on albums (itemno) desc create index clx1 on stock (shipdate) cluster allow reverse scans create unique index incidx on stock (itemno) include (itemname) create index item on stock (itemno) disallow reverse scans collect detailed statistics RENAME INDEX xyz TO pdq Allows to create new index, remove old, rename new name to old name for consistency Page 51

52 Clustering Index DB2 will attempt to store rows with equal or near key values physically close together Improve range searches (e.g. BETWEEN clauses) Only one clustering index may exist for a table Cannot be created on table with APPEND ON Cluster ratio - the degree of data clustering of the index in percentage Higher clustering means rows are ordered on the data pages in index key sequence Cluster factor - a more detailed measurement than the cluster ratio High Cluster Ratio Index Table Low Cluster Ratio Index CREATE INDEX... Page 52

53 Database Administration Multi-Dimensional Clustering (MDC) Multi-Dimensional Clustering Blocks data together related to specified dimensions Index points to block of data - reduces space requirements for indexing Roll-in / roll-out improvements CREATE TABLE MDC1 ( Date DATE, Province CHAR(2), Color VARCHAR(10), YearAndMonth generated as INTEGER(Date)/100,... ) DIMENSIONS ( YearAndMonth, Province, Colour ) Colour Blue Red 23 Region Year Prior to MDC Clustering in one dimension only clustering NOT guaranteed (degrades once page free space is exhausted) 97 Region East East NorthSouth West Year With MDC Clustering guaranteed! Smaller indexes Faster query response Simple definition syntax Fast roll-in & roll-out All records in this block are from the West region and from the year 2000 YearAndMonth AB BC ON QB Province = block 1 Page 53

54 Design Advisor (1/2) Design Advisor can help you design and define suitable indexes: Find the best indexes for a problem query Find the best indexes for a set of queries (a workload), subject to resource limits which are optionally applied Test an index on a workload without having to create the index Can be invoked using either: Control Center or db2advis command Page 54

55 Design Advisor (2/2) Design Advisor Graphical Interface allows you to: Specify the SQL statement / workload for which indexes are to be advised Specify the input file containing one or more SQL statements Specify the maximum space to be used for all recommended indexes in the existing schema Specify the maximum allowable time (in minutes) to complete the operation Save the script to create the recommended objects in outfile Update the catalog statistics Create and schedule a task to create the recommended indexes (if any) Page 55

56 Design Advisor GUI Page 56

57 Design Advisor db2advis Command >>-db2advis---d--database-name > +--w--workload-name-+ +--s--"statement" i--filename '--g ' > > '--a--userid ' '--l--disk-limit-' '-/passwd-' > >< '--t--max-advise-time-' '--h-' '--p-' '--o--outfile-' -d database-name Specifies the database name -w workload-name Specifies the name of the workload for which indexes are to be advised -s "statement Specifies the text of a single SQL statement whose indexes are to be advised -i filename Specifies the name of an input file containing one or more SQL statements -l disk-limit Specifies the maximum space to be used for all recommended indexes in the existing schema -t max-advise-time Specifies the maximum allowable time (in minutes) to complete the operation -o outfile Saves the script to create the recommended objects in outfile Page 57

58 Design Advisor Example 1 Example 1 db2advis -d prototype -s "SELECT * FROM addresses a WHERE a.zip IN ('93213', '98567', '93412') AND (company LIKE 'IBM%' OR company LIKE '%otus')" The utility connects to the PROTOTYPE database, and recommends indexes for the ADDRESSES table Page 58

59 Design Advisor Example 2 Example 2 db2advis -d prototype -w production -l 53 -t 20 The utility connects to the PROTOTYPE database, and recommends indexes that will not exceed 53MB for queries and workload name is "production", the maximum allowable time for finding a solution is 20 minutes Page 59

60 Local Range Partitioned Indexes Easily reorganize partitions based on evolving business needs For example: Roll in and roll out partitions based on new months as needed Easily update territory breakdowns Faster Roll In and Roll Out of data Ability to re-organize specific partitions Quickly detach unwanted partitions for fast data archiving and clean-up Page 60

61 Table Partitioning Improvements: Motivation SQL Automatic Asynchronous Index Cleanup Global Index (order_id) SET INTEGRITY COMMIT CREATE TABLE JUNE Page 61 Jan Feb Mar Apr May ALTER TABLE DETACH PARTITION JAN June LOAD ALTER TABLE ATTACH JUNE

62 Key Improvement: Partitioned Indexes tbsp4 tbsp4 tbsp5 tbsp6 i1 i1 tbsp1 tbsp2 tbsp3 tbsp1 tbsp2 tbsp3 t1.p1 t1.p2 t1.p3 t1.p1 t1.p2 t1.p3 CREATE TABLE t1(c1 INT) INDEX IN tbsp4 PARTITION BY RANGE(a) (STARTING FROM (1) ENDING(34) IN tbsp1, ENDING(67) IN tbsp2, ENDING(100) IN tbsp3) CREATE INDEX i1(c1) CREATE TABLE t1(c1 INT) PARTITION BY RANGE(a) (STARTING FROM (1) ENDING(34) IN tbsp1 INDEX IN tbsp4, ENDING(67) IN tbsp2 INDEX IN tbsp5, ENDING(100) IN tbsp3 INDEX IN tbsp6) CREATE INDEX i1(c1) PARTITIONED Page 62

63 Partitioned Indexes: Benefits Storage Partitioned indexes do not store partition identifier in each key 2 byte savings per RID Each partitioned index will be a fraction of the size of an alternative non-partitiioned index In some cases, may be able to use REGULAR (small RID) tablespaces for indexes Another 2 byte savings per RID Performance Storage savings typically leads to performance savings Less I/O bandwidth and memory consumption Several advantages in utility and DDL processing Page 63

64 WI Department of Information Systems Constraints

65 Constraints Overview Unique Constraints Unique keys Primarykeys Referential Constraints Foreign keys Table Check Constraints Informational Constraints Page 65

66 Unique Constraints Define with CREATE TABLE or ALTER TABLE statement PRIMARY KEY clause or UNIQUE clause Columns must be defined NOT NULL Many unique constraints but at most one primary key per table DB2 uses unique index to enforce unique constraint Existing index might be used Unique index different from unique constraint Can be defined on nullable columns Cannot be parent key for referential constraint Page 66

67 Referential Constraints DEPARTMENT table (Parent table) DEPTNO DEPTNAME MGRNO (Primary key) or unique constraint EMPLOYEE table (Dependent table) EMPNO FIRSTNAME LASTNAME WORKDEPT PHONENO (Primary key) (Foreign key) CREATE TABLE department (deptno INT,... PRIMARY KEY (deptno) CREATE TABLE employee (empno INT,... CONSTRAINT fk_employee_dept FOREIGN KEY (workdept) REFERENCES department ON DELETE NO ACTION ) IN DMS01 Page 67

68 Referential Integrity Rules Insert Rules Rule is implicit when a foreign key is specified Back out insert if not found Delete Rules Restrict Parent row not deleted if dependent rows are found Cascade Deleting row in parent table automatically deletes any related rows in dependent tables No Action (default) Enforces presence of parent row for every child after all other referential constraints applied Set Null Foreign key fields set to null; other columns left unchanged Page 68

69 Referential Integrity Rules Update Rules Restrict Update for parent key will be rejected if row in dependent table matches original values of key No Action (default) Update will be rejected for parent key if there is no matching row in dependent table, meaning if any row in the dependent table does not have a corresponding parent key when the update statement is completed (excluding AFTER triggers), the update is rejected Page 69

70 Table Check Constraints Restricted form of search condition as rule for allowed values Applicable to every row Checked on insert and update You can turn off checking, add the data and then add the constraint, but the table will be placed in CHECK PENDING Page 70

71 Table Check Constraints To modify a constraint you must drop it and create a new constraint CREATE TABLE artists (artno SMALLINT NOT NULL, name VARCHAR(50) WITH DEFAULT 'abc', classification CHAR(1) NOT NULL, bio CLOB(100K) LOGGED, picture BLOB(2M) NOT LOGGED COMPACT ) CONSTRAINT classify CHECK (classification IN ('C','E','P','R')) IN dms01 Page 71

72 Informational Constraints Rules that can be used in query rewrite but are not enforced Standard constraints may result in the overhead for Insert/Update/Delete operations A better alternative if application already verifies data Informational constraint can be used by the SQL compiler but is not enforced by the database manager The SQL compiler includes a rewrite query stage which transforms SQL statements into forms that can be optimized and improve data access path Page 72

73 Informational Constraints Constraint Options ENFORCED The constraint is enforced by the database manager during normal operations such as insert, update, or delete NOT ENFORCED When used, DB2 may return wrong results when any data in the table violates the constraint ENABLE QUERY OPTIMIZATION The constraint can be used for query optimization under appropriate circumstances DISABLE QUERY OPTIMIZATION The constraint cannot be used for query optimization Page 73

74 Informational Constraints Example CREATE TABLE artists ( artno SMALLINT NOT NULL, name VARCHAR(50) WITH DEFAULT 'abc', classification CHAR(1) not null, CONSTRAINT classify CHECK ( classification IN ('C','E','P','R') ) NOT ENFORCED ENABLE QUERY OPTIMIZATION ) INSERT INTO artists VALUES ( 1, 'SMITH', 'C' ), ( 2, 'DONALD', 'P' ), ( 3, 'MAX', 'E' ) ; INSERT INTO artists VALUES ( 4, 'ELLIOT', 'X' ) ; SELECT * FROM artists ; ARTNO NAME CLASSIFICATION SMITH C 2 DONALD P 3 MAX E 4 ELLIOT X 4 record(s) selected. SELECT * FROM artists WHERE classification = 'X' ; ARTNO NAME CLASSIFICATION record(s) selected. DELETE FROM artists WHERE classification = 'X' ; DB20000I The SQL command completed successfully. No results returned because the CHECK constraint is used to optimize the query. If DISABLE QUERY OPTIMIZATION is used, one row will be selected. Page 74