- Overview of system structures
- Containers, table spaces, bufferpools and pages
- Object names
- Tables, views, and indexes
- Stored procedures, triggers and user-defined functions (UDFs)
- Configuration files
- Memory architecture, background processes and threads
- Locking mechanisms
- DB2 pureXML features
- Row compression feature
- Autonomic features
- Downloadable resources
- Related topics
Leverage your Oracle 11g skills to learn DB2 9.7 for Linux, UNIX, and Windows
An earlier developerWorks article talked about how to leverage your MS SQL Server 2000 skills to learn DB2. Is your background in Oracle rather than SQL Server? If so, read on. In this article, we will show you how to use your current knowledge of Oracle 11g to quickly find your footing with DB2 9.7
DB2 9.7 has been delivered with new capabilities to help manage costs and simplify application development. There have been enhancements in several areas such as compression, pureXML, manageability, and performance. This article focuses on making a comparison of DB2 and Oracle fundamental concepts, along with introducing the new features available in DB2 9.7
Note: For the remainder of this article, we will use the term "Oracle" to refer to Oracle 11g and "DB2" to refer to DB2 9.7 for Linux, UNIX, and Windows.
Overview of system structures
To start out, we need to understand the architecture used by Oracle and how it compares with DB2. Figure 1 shows the system structure of Oracle. Compare it with Figure 2, which shows the system structure of DB2. Refer back to these figures for understanding as you read the article.
Figure 1. Oracle system structure on Linux, UNIX, and Windows
Figure 2. DB2 on Linux, UNIX, and Windows system structure
The concept of an instance is similar in both Oracle and DB2. In both cases an instance is a combination of background processes and shared memory. The major difference between the two is that in Oracle there can only be one database per instance, while in DB2 multiple databases can share an instance.
Because there is a one-to-one correspondence between database and instance, in Oracle you create an instance implicitly by creating a database with the CREATE DATABASE command. Alternatively, to create an Oracle instance on your machine you can use the Database Configuration Assistant, or you can use the ORADIM utility, available with Oracle 9i with the NEW option. You must also provide certain information including a System Identifier (SID) or a service name, instance password, maximum number of users, startup mode, and so on. Similarly, to delete the instance the ORADIM utility can be used with the DELETE option. You would need to pass in either the SID or the service name. There is no default instance created with a fresh install of Oracle unless you create a new database during the installation process.
In DB2, after installing the product on the Windows platform, the instance
"DB2" is created by default. In Linux and UNIX, the default instance name
is called "db2inst1". To create another instance in the same machine, you
simply execute the command
db2icrt <instance name>.
Figure 3 shows one instance called DB2_01 created with the db2icrt command from the DB2 Control Center GUI.
Figure 3. DB2 Control Center GUI showing DB2 instances
To reference a given DB2 instance from a command line interface, use the
environment variable DB2INSTANCE. This variable lets you specify the
current active instance to which all commands would apply. For example, if
DB2INSTANCE is set to PROD, and then you issue the command
create database MYDB1, you will create a database associated
with instance PROD. If you wanted to create this database on instance DB2
instead, then you would first have to change the value of the DB2INSTANCE
variable to DB2. This is similar to the ORACLE_SID (System Identifier)
which is also used when users want to switch between instances.
Another easy way to identify the instance you want to work with is by using
the DB2 Control Center GUI as shown in Figure 3. To
see an entry for the new instance in this tool, you may have to add the
instance to the GUI by right clicking on Instances and
choosing Add. To drop an instance in DB2, you can execute
db2idrop <instance name>.
In summary, in Oracle the Database Configuration Assistant can be used to create, modify, start, stop and delete the instance, while in DB2 the Control Center GUI can be used for a similar purpose. Also an Oracle instance can only have a one-to-one relationship with a database, while in DB2 this is not the case. Multiple databases can exist and be concurrently used in a DB2 instance.
In Oracle, a database can either be created manually using the CREATE DATABASE command or by using the Database Configuration Assistant. Creating the database manually requires that a series of steps be followed, including setting OS variables, preparing the parameter file, and creating a password file before the CREATE DATABASE command can be executed.
Metadata information is stored and managed by the Data Dictionary, which is composed of base tables, and corresponding views. The base tables are created automatically during database creation, and the views are constructed by running the catalog.sql and catproc.sql scripts.
The Oracle database is therefore seen as a collection of three file types:
- Data File: Contains actual data, physical implementation of the database. (Similar to Containers in DB2)
- Redo File: This is equivalent to Transaction Logs in DB2.
- Control File: Contains information to maintain and verify database integrity.
In DB2, an instance can contain several databases, as shown in Figure 2. Each database is a truly closed and independent unit. Each database has its own catalog table space, temporary table space, and user table space which are created by default upon successful database creation. DB2 does contain a binary file known as the system database directory that contains entries of all the databases you can connect from your DB2 machine. This directory is kept at the instance level.
When an instance is created, no databases are created by default. You need to explicitly create a database using the create database command. You can also create a database using the Control Center as shown in Figures 4 and 5.
Figure 4. Creating a DB2 database using the Control Center GUI
Figure 5. Creating a DB2 database using the Control Center GUI (continued)
In Figure 5, you can also see what happens when you click Show Command. All DB2 Control Center GUI screens let you see the SQL statement or command that is actually executed in the background. These commands can be saved in scripts for execution at a later time, or can be copied and executed from the Command Line Processor (CLP) tool or Command Center GUI tool. These tools are equivalent to Oracle's SQL*Plus and iSQL *Plus respectively.
A DB2 database can be dropped by using the 'DROP DATABASE' command or from the DB2 Control Center GUI. In Oracle there is no such command; a database is deleted by deleting all the associated datafiles.
Databases within an instance normally don't interact with each other. However, if your application needs to interact with more than one database, this requirement can be supported by enabling federation support. See the resources section for an article about federation.
Containers, table spaces, bufferpools and pages
In Oracle, data is physically stored in files called Data Files. This is similar to DB2's containers where data is physically stored. Every Oracle database contains a table space named SYSTEM, which Oracle creates automatically when the database is created. Other table spaces for user, temporary and index data need to be created after the database has been created, and a user needs to be assigned to these table spaces before they can be used.
In DB2, a table space is the logical object used as a layer between logical tables and physical containers. When you create a table space, you can associate it with a specific buffer pool (database cache) as well as to specific containers. This gives you flexibility in managing performance. For example, if you have a "hot" table, you can define it in its own table space associated with its own buffer pool. This helps ensure the data for this table is continuously cached in memory.
In DB2 three default table spaces are created automatically upon database creation when using default values for the CREATE DATABASE command. Table 1 describes the default DB2 table spaces:
Table 1. DB2 table spaces created by default when a database is created with default values
|Table space Name||Description|
|SYSCATSPACE||Catalog table space containing metadata.|
|TEMPSPACE1||System temporary table space used to perform operations such as joins and sorts. The name of this table space can be changed.|
|USERSPACE1||This table space is optional and can be used to store user tables when a table space is not explicitly indicated at table creation time.|
Because databases in DB2 are independent units, table spaces are not shared across databases. Since they are only known within a database, two different databases can have table spaces of the same name. You can see this in Figure 2 where database MYDB1 has a table space named MYTBLS and database MYDB2 has a table space of the same name.
DB2 table spaces can be classified as SMS (system-managed spaces) or DMS (database-managed spaces). SMS table spaces are managed by the operating system and can only be directories. They grow automatically as needed; thus SMS provides good performance with minimum administration. DMS table spaces are managed by DB2, and can be files or raw devices. This type of table space allows for best performance, but some administration is required. For example, you need to specify ahead of time the amount of space you want to allocate for the table space, because growth is not automatic.
Oracle does not have the SMS concept for its storage model but its data files are similar to DB2 DMS table spaces. That is, you can increase a database size by either adding a datafile to the table space, by increasing the size of the data file, or by adding a new table space.
Table 2 below shows how Oracle databases or table spaces map to DB2 databases or table spaces.
Table 2. How Oracle databases map to DB2 databases and table spaces
|Oracle database or table space||DB2 database or table space|
|SYSTEM is the table space that holds the catalog (Data Dictionary) information.||SYSCATSPACE (catalog table space); as in Oracle, this information is kept only at the database level.|
|Data Dictionary (contains metadata in form of tables and views) and resides inside the SYSTEM table space.||System Catalog Tables (identified by the SYSIBM schema) and system views (identified by the SYSCAT OR SYSSTAT schema) and reside inside the SYSCATSPACE table space|
|SCOTT database||SAMPLE database|
|TEMP table space||System Temporary table space (By default it is called tempspace1)|
|UNDO table space||N/A|
|USER table space||User table space. By default, USERSPACE1 is normally created after database creation|
As indicated earlier, Oracle's data buffer concept is equivalent to DB2's bufferpool; however, DB2 allows for multiple bufferpools to exist. There is no predefined number of bufferpools that you can create, and they can have any name.
The concept of an Oracle block is most similar to DB2's page. A DB2 page can have a size of 4k, 8k, 16k or 32k. A table row must fit in only one page; it cannot span to other pages as in Oracle.
An Oracle object name takes the following form:
In DB2, objects also have a two-part structure:
As in Oracle, the DB2 schema name is used to group objects logically. An important difference, however, is that in DB2, a schema name does not have to match to a user id. Any user with a privilege called IMPLICIT_SCHEMA can create an object using a non-existing schema. For example, suppose that "Peter" has IMPLICIT_SCHEMA privilege and executes this command:
CREATE TABLE WORLD.TABLEA (lastname char(10))
In this case, the table WORLD.TABLEA is created, where WORLD is the newly created schema. If Peter had not explicitly indicated the schema, then the table PETER.TABLEA would have been created, because the connection ID is used by default.
In DB2 you always connect to a database before issuing database specific commands; therefore, under this architecture, object names need not include the database name.
Tables, views, and indexes
Tables, views, and indexes are basically the same in both Oracle and DB2.
DB2 provides a utility called Design Advisor which you can use to recommend indexes for a particular query or workload. You can invoke the Design Advisor from the DB2 Control Center or from the DB2 CLP using the db2advis command. In DB2, indexes are directly tied to the table definition. For example, when using DMS table spaces you can specify in which table space the indexes can reside as follows:
CREATE TABLE mytable (col1 integer, col2 char(10)) in tbls1 index in tbls2
The above example shows that the data for the table will be stored in table space 'tbls1', while the index pages will be stored in table space 'tbls2'. This is in contrast to Oracle syntax where the CREATE INDEX statement provides an option to specify in which table space the index will reside.
Also, once an index has been created in DB2, you cannot alter any clause of the index definition. You would need to drop the index and create it again in order to implement changes.
As in Oracle, DB2 tables, views and indexes in different databases can have the same names. Tables and views within the same database must have distinct names, but creating an index with the same name as either an existing table or view is allowed.
Stored procedures, triggers and user-defined functions (UDFs)
DB2 stored procedures can be written in any language supported by the DB2 precompilers, including Java, C, C++, REXX, Fortran, and COBOL. However, the recommended language to use is the SQL Procedural Language (SQL PL), which is fairly similar to Oracle's PL/SQL. SQL PL stored procedures are easy to create and have good performance. DB2 stored procedure development also supports SQLJ and Java using JDBC drivers type 2 and 4. Type 3 is discontinued.
Development of triggers and functions can use inline SQL/PL, a subset of SQL PL. The Data Studio tool can be used to easily create, build, debug and deploy DB2 stored procedures and user-defined functions.
Traditionally Oracle stores all session and system-related parameters in a text file, normally referred to as initSID.ora. However due to the non-persistent nature of this text file, starting from Oracle 9i, Oracle introduced Server Parameter File (SPFILE), which is a binary parameter file stored on the server. This persists across instance shutdown and startup. The initSID.ora file is still used, however, when an SPFILE is not available. Prior to the introduction of SPFILE, any ALTER SYSTEM and ALTER SESSION commands that affected parameters would only persist during that instance or that session. The DBA would have had to manually modify the initSID.ora text file whenever a rebound of a database instance was intended. Network access configurations are typically stored in listener.ora for listener and tnsnames.ora for client access.
With DB2, configuration parameters are stored at the instance level, known as the database manager configuration file, and at the database level, known as the database configuration file. Most of these parameters can be changed dynamically, that is, there is no need to stop and restart the instance or force all connections from a database before the change to the parameter value takes effect. The files where DB2 stores its configuration information cannot be edited directly.
If you would like to manually change a specific database manager parameter
from the CLP, use the command
DBM CFG USING <parameter name> <new value>.
If you would like to manually change a specific database parameter from the
CLP, use the command
UPDATE DB CFG FOR
<database name> USING <parameter name> <new value>.
These commands would be the equivalent to Oracle's ALTER SYSTEM and ALTER SESSION. Alternatively, using the Control Center, you can review and change values for these parameters; if you right click on a given instance and choose Configure Parameters, you will see the window shown in Figure 6.
Figure 6. DB2 Database Manager configuration parameters (instance level)
At the database level, right clicking on a given database, and choosing Configure Parameters displays the window shown in Figure 7.
Figure 7. Database configuration parameters (database level)
DB2 provides many parameters you can use to configure your system. However,
if you would like an easy way to configure the system automatically, use
autoconfigure command (or the Configuration Advisor GUI)
which sets the database manager and database configuration parameters to
optimal values based on some information you provide. Figure 8 shows the Configuration Advisor.
Figure 8. - DB2 Configuration Advisor
In addition to configuration files, DB2 also uses DB2 Registry variables normally for platform-specific configurations. Note, the DB2 Registry variables have no relationship whatsoever to the Windows registry. Use the command db2set to review and change these variables.
Connectivity (network access) information is stored in the system database directory, local database directory and node directory. These are binary files and can only be modified with the CATALOG and UNCATALOG commands.
Memory architecture, background processes and threads
Next we'll look at the memory architecture, background processes and threads, and will compare and contrast the way they are used in Oracle and DB2.
Figure 9: Oracle memory architecture and background processes
The System Global Area (SGA) in Oracle is a group of shared memory areas that stores information for the instance. Examples include statement cache, redo log buffers, and data buffer cache. The Program Global Area (PGA) and the User Global Area (UGA) shared memory areas contain data and control information for server processes and user sessions.
Oracle supports multiple instances within the same machine but background processes are not shared. For example, three instances in one machine will require three set of background processes. Therefore it is generally recommended to have one database, one instance and multiple schemas within the same machine.
Figure 10: DB2 Memory architecture and background processes and threads
Both DB2 and Oracle use shared memory areas, but DB2's memory architecture is implemented in a slightly different way than Oracle's. Since a DB2 instance can contain more than one database, two levels of configuration exist. As we mentioned in the previous section, instance level configuration can be done in the DBM CFG file while the database level configuration is done in the DB CFG file. Configuration parameters at both levels can be adjusted to tune memory usage. The section below provides a bit more detail on DB2's memory structures and different background processes.
Unlike Oracle where memory is allocated to both the instance and the database upon startup, DB2 allocates memory at different levels. This is primarily due to the fact that a DB2 instance can contain multiple databases. There are three main memory structures in DB2:
- Instance shared memory: This refers to the database
manager global shared memory, which is allocated when the instance is
started using the
db2startcommand, and remains allocated until a
db2stopcommand is issued to stop the instance.
- Database shared memory: This refers to the database global memory, which is allocated when the database is activated or connected to for the first time. Memory allocated includes buffer pools, locklist, database heap, utility heap, package cache and catalog cache.
- Application shared memory: This refers to the memory allocated when an application connects to a database and is used by agents that do the work requested by the connected clients. Each application connected to the database has memory allocated to it; therefore accurate configuration of the parameters affecting the application shared memory becomes crucial.
The DB2 database server must perform many different tasks, such as processing database application requests or ensuring that log records are written out to disk. Each task is typically performed by a separate engine dispatchable unit (EDU). In DB2 for Windows, Linux and UNIX , server activities are conducted in the form of threads. DB2 threads and processes operate in the following levels. Note that this is a non-exhaustive list but we do highlight the important threads and processes:
- Instance Level: These are processes and threads that are initialized
when an instance is started:
- DB2 Daemon Spawner (db2gds): Global daemon processor started for each instance (only in UNIX)
- DB2 System Controller (db2sysc): Main DB2 process
- DB2 Watchdog (db2wdog): Parent process for all other processes
- DB2 Format Log (db2fmtlg): Similar to the ARCn process in Oracle, pre-allocates log files in the log path
- Autonomic computing daemon (db2acd): Hosts the health monitor, automatic maintenance utilities, and the administrative task scheduler. This process was formerly known as db2hmon.
- Database Level: These are processes that are initialized when a
connection is made to a database.
- DB2 Log Reader (db2loggr): Similar to the subset of Oracle's PMON process. This process reads log files during rollback, restart recovery and roll forward.
- DB2 Log Writer (db2loggw): Flushes log from log buffer to the transaction log files on disk. Equivalent to LGWR process in Oracle.
- DB2 Page Cleaner (db2pclnr): Equivalent to DBWR process in Oracle, this process cleans the buffer pool before pages from the disk are moved into the BP.
- DB2 Prefetcher (db2pfchr): Retrieves pages from disk and places into the buffer pool before it is needed.
- DB2 Deadlock Detector (db2dlock): Deadlock detector process.
- DB2 Self-Tuning Memory Manager (db2stmm): for the autonomic self-tuning memory management feature.
- Application Level: Each application connecting to the database would
have its own share of application-level background processes
associated with it. These are as follows:
- DB2 Communication Manager (db2ipccm): Inter-process communication process for each locally connected client.
- DB2 TCP Manager (db2tcpcm): TCP communication manager process for remote clients connecting using TCP/IP.
- DB2 Coordinating Agent (db2agent): Thread that handles all requests on behalf of an application.
- DB2 Pooled Gateway Agent (db2agntgp and db2agntgp): An agent pooled on a remote database and local database respectively.
For a comprehensive explanation of DB2's processes please refer to the article 'Everything you wanted to know about DB2 processes'.
Locking in Oracle can be either manual or automatic. The Oracle Lock
Manager can implicitly lock table data at the row level, or default locks
can be overridden at transaction or session level using the following SQL
1. SET TRANSACTION ISOLATION LEVEL
2. LOCK TABLE
3. SELECT FOR UPDATE
Oracle supports a mechanism called Multi-Version Read Consistency which is implemented by undo data in the undo segments.
DB2 implements ANSI standard Isolation levels such as Uncommitted Read,
Cursor stability, Read stability and Repeatable Read. A user will only see
committed data unless the Uncommitted Read isolation level is used. Row
locks are acquired implicitly according to the isolation level. Database
objects that are lockable are table spaces, tables and rows, however, only
tables and table spaces can be explicitly locked. The
LOCK TABLE command can be used to lock a table instead of
using the default row locking.
Unlike Oracle, in DB2, locks are stored in the memory and not in data pages. The LOCKLIST database configuration parameter can be used to configure the memory available for locks, while the MAXLOCKS configuration parameter defines the maximum amount of memory for a particular application's locks.
In DB2 9.7 lock event reporting has been enhanced. The new lock event monitor collects information about lock timeouts, deadlocks, and lock waits that are more than a specified duration. This data can be accessed from either a XML document, from database tables or using a java based tool (db2evmonfmt) to read from an xml or text document.
A new database configuration parameter
cur_commit has been
introduced which basically allows only committed data to be returned, as
was the case previously, but now readers do not wait for writers to
release row locks. Instead, readers return data that is based on the
currently committed version; that is, data prior to the start of the write
Both Oracle and DB2 are secure databases with basic and advanced security features. In Oracle, there are 4 different user authentication methods, as follows:
- Database: The database performs both identification and authentication of users.
- External: The operating system or network service performs authentication
- Global authentication and authorization: The user is authenticated globally by SSL
- Proxy authentication and authorization: The middle-tier server performs authentication.
The authentication method is specified when creating the user using the
CREATE USER command. There are several Data Dictionary views
that contain information about these users.
In DB2, users do not exist within the database, but are rather managed by the operating system. No database login information is kept in any database table. Any operating system user can potentially access DB2; however, unless they have been granted a given DB2 authority or privilege, there is not much they can do. Granting and revoking authorities and privileges can easily be handled through the Control Center GUI. You first may have to add a user or group to the Control Center from the available operating system users or groups.
Also in DB2 the term "roles" is not used; instead, DB2 uses the term "authorities", which are similar to Oracle's database roles, whereby privileges are granted to certain groups or users. The authorities supported with DB2 are: SYSADM, SYSCTRL, SYSMAINT, SYSMON, SECADM, DBADM and LOAD.
The SYSADM, SYSCTRL and SYSMAINT authorities cannot be granted using the GRANT SQL statement. These special authorities can only be set by modifying the specific database manager configuration parameters.
DB2 also uses the term "privilege" which is similar to Oracle's system and schema object privileges. There are database privileges (connect, createtab, and so on) and database object privileges (schema, table, view, and so on). Figure 11 shows DB2 security information obtained from the Control Center GUI. Most of the tabs shown on the Change User window correspond to the privileges supported by DB2.
Figure 11. DB2 security
Authentication in DB2 does not only involve encrypting user names and passwords but also allows for encryption of data as it travels across the network between the clients and the server. The location of the authentication process is determined by the value of the database manager configuration parameter AUTHENTICATION.
The following are the valid options for enabling authentication for DB2:
- SERVER_ENCRYPT - This value specifies that the authentication occurs on the server. The user id and password specified during connection are encrypted and sent to the server where they are compared with the user and password on the server side. If the match is successful, the user is allowed to access the database.
- KRB_SERVER_ENCRYPT - This specifies that the server accepts KERBEROS authentication or encrypted SERVER authentication schemes.
- DATA_ENCRYPT - This setting specifies that the server allows for SERVER authentication, and also that the data traveling over the network between the client and server is encrypted.
- DATA_ENCRYPT_CMP - This specifies that the server accepts encrypted SERVER authentication schemes and the encryption of user data. This authentication type allows for compatibility with down-level products not supporting the DATA_ENCRYPT authentication type.
- GSS_SERVER_ENCRYPT - This specifies that the server accepts GSS API-based plug-in authentication or encrypted server authentication schemes.
To update the AUTHENTICATION instance parameter, for example to a value of DATA_ENCRYPT, use the commands shown below:
Listing 1. Updating AUTHENTICATION instance parameter
UPDATE DBM CFG USING AUTHENTICATION DATA_ENCRYPT db2stop db2start
DB2 further extends security by offering a Label Based Access Control (LBAC) mechanism. The LBAC feature provides for greater granularity for controlling read and write access to individual rows and table columns. A security administrator role (SECADM) has been provided in DB2 which is needed to manipulate LBAC objects. Users attempting to access an object must have its security label granted to them. When there's a match, access is permitted; without a match, access is denied.
There are other aspects of database security beyond privileges and authorities. Briefly, here are some of the differences and similarities between Oracle and DB2:
User authentication and authorization
Oracle uses an encrypted password stored in the dictionary once a user is created. DB2 supports passwords for user authentication and uses the underlying operating user for authentication. Both Oracle and DB2 support LDAP (Oracle Internet Directory and IBM Directory Server). Both Oracle and DB2 support single sign on (SSO).
Oracle supports data encryption where sensitive data such as credit card numbers and some highly sensitive business data can be encrypted. You have the following options for encrypting DB2 data in storage:
- You can use the encryption and decryption built-in functions ENCRYPT, DECRYPT_BIN, DECRYPT_CHAR, and GETHINT to encrypt your data within database tables.
- You can use IBM Database Encryption Expert to encrypt the underlying operating system data and backup files.
- If you are running a DB2 Enterprise Server Edition system on the AI operating system, and you are interested in file-level encryption only, you can use encrypted file system (EFS) to encrypt your operating system data and backup files.
Oracle provides network encryption with its Oracle Advanced Security. Oracle uses DES, 3DES and RC4 industry standard encryptions. To encrypt data in-transit between clients and DB2 databases, you can use the DATA_ENCRYPT authentication type, or, the DB2 database system support of Secure Sockets Layer (SSL).
Oracle allows you to audit trail users and objects. Log miner can also be used for investigating and analyzing suspicious queries. DB2 provides a similar auditing facility. The db2audit utility can be used for this purpose.
This section compares Oracle's support for XML to DB2's. Shipped with Oracle 9i Release 2, Oracle XML DB feature provided a way to manage XML storage, retrieval and schema by defining XMLTYPE tables and columns which are stored as CLOB or shredded (decomposed) to small pieces into relational tables. Oracle 10g came with some improvements to manage XML documents; for example, changes in schema can be dynamically reflected by mapping existing data without the need to re-import. In Oracle 11G, a third kind of XML support called binary XML was introduced. So Oracle now has the following ways of storing XML data:
- Unstructured storage using CLOBs (also know as schema-less storage)
- Structured storage which maps XML to object relational
- Binary XML storage
DB2's pureXML technology, stores XML documents natively, that is, internally in tree format. It also allows for the use of SQL with XML extensions, Xquery and Xpath to access relational and XML data. Storing XML documents natively is a better approach, and IBM's research indicates that there is a better performance for search and retrieval of XML documents, and reduction in the lines of codes for certain programs. Support for the DB2 XML data type is available in the DB2 Control Center, command line processor, IBM Data Studio, and IBM Database Add-Ins for Microsoft Visual Studio.
In order to use the pureXML feature in your database, create the database as UNICODE (for example using codeset UTF-8). Failure to create a UNICODE database prior to creating a table will result in an error as shown below:
SQL1239N XML features can only be used in a Unicode database with a single database partition. SQLSTATE=42997
DB2 stores relational data as in previous versions. However, XML data is stored in hierarchical format (as a tree using the Xquery Data Model, XDM). There is tight integration between both XML and relational services. To store XML documents, a user needs to create a table, and specify a column to use a new data type, XML, as shown in the example below.
Listing 2. Create a DB2 table with XML datatype
CREATE TABLE T (ID INT PRIMARY KEY NOT NULL, COMMENT VARCHAR(1000) NOT NULL, DOCUMENT XML) in ttspace compress yes@
As XML documents are stored in parsed hierarchical format natively in the XQuery Data Model (XDM), there is no need for translation or mapping; the format used to store XML documents is the format used to process it. This allows for greater performance.
Utilities such as backup, restore and import apply to tables with XML columns the same as in any other tables. XML data can be inserted into the XML column either by using the INSERT statement or by using the DB2 IMPORT utility. Prior to importing XML documents received from third parties, it is a good idea to validate these documents against a pre-defined XML schema. To register against an XML schema, DBAs are required to issue the REGISTER XML SCHEMA command, ended with COMPLETE XML SCHEMA to complete the registration process. DB2 also supports indexes to be created on a subset of an XML document, or on the entire document. An XPATH needs to be specified when creating an index which would point to that particular element/attribute to be indexed.
With DB2 you now have four ways to access relational and XML data:
- Plain SQL (no XQuery involved)
- SQL/XML, that is, XQuery embedded in SQL
- XQuery as a stand-alone language (no SQL involved)
- Xquery with embedded SQL
In DB2 9.7 additional XML features were included such as support for the XML type in User Defined Funcions. Important enhancements in DB2 9.7 are that the LOAD utility can now be used to load XML data. Also, On partitioned tables, indexes over XML data that you create with DB2 V9.7 or earlier are nonpartitioned. Starting in DB2 Version 9.7 Fix Pack 1, you can create an index over XML data on a partitioned table as either partitioned or nonpartitioned.
Oracle Partitioning offers several partitioning strategies that control how the database places data into partitions. The basic strategies are range, list, and hash partitioning.
DB2's table partitioning (a.k.a Range partitioning), is similar to Oracle's partitioning. It basically allows a single logical table to be broken up into multiple physical storage objects across one or more table spaces. Each of the objects would correspond to a 'partition' and would allow each table space to contain a range of data which can be accessed very easily.
In DB2 there are several ways to partition your data, and you can apply these methods simultaneously on the same data. To avoid confusion, here is a short explanation about the different ways to provide this partitioning:
- DATABASE PARTITIONING - distributing data by key hash across logical nodes of the database (DPF).
- RANGE/TABLE PARTITIONING (Available with DB2 9) - splitting data by key range over multiple physical objects within a logical database partition.
- MULTI DIMENSIONAL CLUSTERING (MDC) - organizing data in table (or range of a table) by multiple key values.
In Version 9.7, you can have indexes that refer to rows of data across all partitions in a data partitioned table (known as nonpartitioned indexes), or you can have the index itself partitioned such that each data partition has an associated index partition. You can have both nonpartitioned and partitioned indexes for partitioned tables.
The following example creates a table customer where rows with l_shipdate >= '01/01/2006' and l_shipdate <= '03/31/2006' are stored in table space ts1, rows with l_shipdate >= '04/01/2006' and l_shipdate <= '06/30/2006' are in table space ts2, etc. More exhaustive explanation can be obtained from developerWorks article, Table partitioning in DB2 9.
Listing 3. Range partitoning a table
CREATE TABLE customer (l_shipdate, l_name CHAR(30)) IN ts1, ts2, ts3, ts4, ts5 PARTITION BY RANGE(l_shipdate) (STARTING FROM ('01/01/2006') ENDING AT ('12/31/2006') EVERY (3 MONTHS))
A comparison of Oracle and DB2 partitioning terminology can be found in Table 1 of the article Migrate from Oracle or Sybase to DB2 in weeks.
There are three compression features provided by Oracle; index, table and row level compression. Improper planning of these features may have adverse effect in performance.
Oracle introduced index compression since version 8i. Indexes that can be compressed are bitmap, btree and index organized tables. To use index compression is straightforward; for example, to create an index with the compression feature, use:
Listing 4. Create index with compression
CREATE INDEX ord_customer_ix_demo ON orders (customer_id, sales_rep_id) COMPRESS 1;
For indexes that were not created initially with compression, you can turn them to compression by altering them. Listing 5 shows an example of how you can alter an index to include compression.
Listing 5. Alter index with compression
alter index ord_customer_ix_demo rebuild compress
Currently, Oracle doesn't provide any automated advisor to pinpoint which indexes should be compressed. Most of the benefits gained by index compression requires proper planning by seasoned DBAs who with intimate knowledge of Oracle CBO.
Table compression on the other hand, was introduced in Oracle 9i release 2. It can be used to compress entire tables, table partitions, and materialized views. The compression can be applied to all partitions or some partitions. Though table compression works for a non-partitioned table, its usage for non-partitioned tables in OLTP workloads may not be desirable as insert and update performance may suffer. In Oracle table compression, duplicate values are removed in a database block, and information is stored to recreate the uncompressed data within the block.
The following example shows how to create partition table with compression.
Listing 6. Create table with compression
CREATE TABLE costs_demo ( prod_id NUMBER(6), time_id DATE, unit_cost NUMBER(10,2), unit_price NUMBER(10,2)) PARTITION BY RANGE (time_id) (PARTITION costs_old VALUES LESS THAN (TO_DATE('01-JAN-2003', 'DD-MON-YYYY')) COMPRESS, PARTITION costs_q1_2003 VALUES LESS THAN (TO_DATE('01-APR-2003', 'DD-MON-YYYY')), PARTITION costs_q2_2003 VALUES LESS THAN (TO_DATE('01-JUN-2003', 'DD-MON-YYYY')), PARTITION costs_recent VALUES LESS THAN (MAXVALUE));
To turn a table into a compressed table, use alter table <table name> move compress. A compressed table, however, does not allow adding or dropping of columns.
In DB2 9.7, row compression, also referred to as deep compression, compresses data rows by replacing patterns of values that repeat across rows with shorter symbol strings. Of the various data compression techniques available in DB2 9.7, row compression offers the biggest possibilities for storage savings. Row compression requires the creation of a dictionary which stores a mapping between repetitive patterns or entries and numeric keys. The compression algorithm is intelligent enough not to compress rows that will not realize in any significant disk-space saving.
DB2's row compression, unlike Oracle's key compression, does not require keys to be specified.
Compression is enabled at individual table level via the CREATE TABLE or ALTER TABLE commands. For example:
Listing 7. Create/Alter table with COMPRESSION YES
CREATE TABLE Sales COMPRESS YES ALTER TABLE Sales COMPRESS YES
To achieve the same effect using the DB2 Control Center, during the column definition (second step in the table creation wizard), ensure that you choose the checkbox, Store table data in a compressed format located at the bottom of the panel (as the following diagram shows).
Figure 12. DB2 Control Center - Creating table with compression
The table dictionary is built only when a REORG is performed, after which the data in the table can be compressed. The dictionary gets updated for every subsequent REORG operation. The compressed data is kept both on disk and memory and DB2 also compresses user data stored in log files, thereby reducing log file size.
Note that each partition of a partitioned table can have a different compression dictionaries and each partition of a table in DPF can have a different compression dictionaries.
Besides, data row compression, other compression mechanisms offered by DB2 9.7 include:
- NULL and Default Value Compression (V8 GA): Compression for null, zero length data in variable length columns and system default value.
- Multidimensional Clustering (V8 GA): Implements a form of index compression by using block indexes, one index entry for thousands of records.
- Database Backup Compression (V8 FP4): Compression to result in smaller backup images.
- XML Parsing
There are a few tuning improvements provided in Oracle 11g. Oracle automated the following tuning areas:
- Redo Logfile Sizing Advisor - This feature recommends the optimal size of redo log files in order to avoid excessive disk I/Os due to frequent checkpointing.
- Automatic Checkpoint Tuning - The Oracle Database can now self-tune checkpointing to achieve good recovery times with low impact on normal throughput. You no longer have to set any checkpoint-related parameters.
- Automatic Shared Memory Tuning - Automatic Shared Memory Tuning automates the configuration of System Global Area (SGA) memory-related parameters (buffer cache, shared pool) through self-tuning algorithms. It simplifies database configuration, ensures most efficient utilization of available memory and improves performance.
- Transaction Rollback and Recovery Monitoring - This feature enables you estimate how long it will take to roll back a transaction. You can also monitor the progress of transactions being recovered and estimate the average speed of transaction recovery.
- SQL Tuning - The SQL Tuning Advisor automatically tunes high cost SQL statements.
- Automatic Storage manager - The automatic storage manager (ASM) simplifies the managment or Oracle related files.
Oracle also provides some advisors such as segment and undo advisors. The segment advisor is based on the level of space fragmentation within an object, and as result, gives advice on whether an object is a good candidate for the new online shrink operation. As well, this advisor give reports on the historical growth trend of segments, and proved to be especially informative for capacity planning.
Undo Advisor, on the other hand, helps administrators to make the right judgement in sizing the undo table space in both flash back and non-flashback features. It advises administrators in setting UNDO_RETENTION appropriately to avoid the age-old 'snapshot too old' problem.
DB2 9.7 has several autonomic features to assist in managing its environment that are self-configuring, self-healing, self-optimizing, and self-protecting. By sensing and responding to situations that occur, autonomic computing shifts the burden of managing a computing environment from database administrators to technology.
Automatic memory tuning
DB2 9.7 features a self-tuning memory feature called the Self Tuning Memory Manager that simplifies the task of memory configuration by automatically setting values for several memory configuration parameters. When enabled, the automatic tuner acting as dispatcher will figure out the memory resources available and distribute them to several memory consumers for the database dynamically. Self-tuning memory only applies to single partition databases.
With the AUTOCONFIGURE command you can calculate and displays initial values for the buffer pool size, database configuration and database manager configuration parameters, with the option of applying these recommended values.
Automatic storage management
Automatic storage automatically grows the size of your database across disk and file systems and since it automatically grows the size of the database, it removes the need to managing storage containers by DBAs. When you create databases in DB2 9.7 the automatic storage management feature is enabled by default.
DB2 9.7 has automatic maintenance capabilities that are used for automatically performing maintenance functions such as:
- Automatic database backups which provide the capability of a full database backup performed as needed.
- Automatic statistics collection. DB2 determines which statistics are required and need to be updated and then automatically executes the RUNSTATS utility in the background.
- Automatic table and index reorganisation. DB2 determines if a table or index reorganization by periodically checking tables and indexes that have had their statistics updated, and schedules such operations whenever they are necessary.
We shall look at tools from different areas such as database creation and maintenance, network, administration GUI, performance tuning, data movement and backup-recovery tools. Figure 13 shows the DB2 9.7 GUI tools.
Figure 13. DB2 9.7 GUI Tools
Let's take a look at how similar tasks are performed in Oracle and DB2 9.7.
Database Creation and Maintenance
Oracle provides the Database Configuration Assistant (dbca) as the GUI tool to create databases. For database maintenance, Oracle provides the Oracle Enterprise Manager. DB2 databases can be created and maintained from the DB2 Control Center.
Oracle provides the Network Configuration Assistant (netca) for network configuration. Alternatively, you can use the Oracle Network Manager to configure service naming, listener, profile and Oracle name servers. DB2 uses the CATALOG command to catalog nodes and databases. Cataloging can also be done using either the DB2 command line or DB2 Configuration Assistant GUI.
The Oracle Enterprise Manager provides a wide range of administrative capabilities for the day-to-day tasks of administrators. DB2 Control Center provides similar functions as Oracle Enterprise Manager. Apart from the DB2 Control Center, you can also use the DB2 command line processor to issue DDL and DML statements. This utility is similar to the Oracle SQLPLUS utility. Figure 14 shows the DB2 command line processor.
Figure 14. DB2 Command Line Processor
Commands can also be issued from the Command Center, shown in Figure 15.
Figure 15. Command Center GUI (GUI version of DB2 Command Line Processor)
Oracle Enterprise Manager comes with Change Management pack, Tuning Pack, and Diagnostic Pack. DB2 provides the Activity Monitor, Event Analyzer, Health Center, Indoubt Transaction Manager and Memory Visualizer as GUI tools for performance tuning tasks.
Oracle provides SQL Loader (sqlldr) for loading data in delimited text format. Import (imp) and export (exp) can be used to perform logical import and export. DB2 provides similar import, export and load utilities . For cross platform data movement, DB2 provides the db2move utility.
Backup and Recovery
Oracle provides Recovery Manager as an option for hot backup. Backing up databases in DB2 can be accomplished using the backup command or the DB2 Control Center.
Oracle 11g Enteprise Manager comes with new performance overview charts. The enhanced Oracle Enterprise Manager HTML interface provides a central point of access to all database performance-related statistics and facilitates complete monitoring and diagnostics.
Apart from interfaces that are shipped with DB2 9.7 there is a no-charge application development tool, based on the Eclipse framework called tbe IBM Data Studio (Data Stuidio). Data Stuidio is a one-stop center for creating, editing, debugging, deploying, and testing DB2 stored procedures and user-defined functions. You can also use the Data Stuidio to develop SQLJ applications, and create, edit, and run SQL statements and XML queries.
You can download IBM Data Studio from the Developerworks web site.
For examples and features, refer to the IBM Data Studio Features and Benefits web page. article in developerWorks.
In this article we've introduced you to DB2 9.7 for Linux, UNIX and Windows by using your current knowledge of Oracle 11g as leverage. We briefly described the DB2 architecture, background processes, memory model, security, tools, and so on. There are many similarities between Oracle and DB2 9, and we have pointed out some of the differences so you can use your current knowledge to become successful with DB2 9.7.
Table 3 summarizes the differences and similarities between Oracle and DB2 that we've discussed.
Table 3. Summary of Oracle vs. DB2 concepts
|Instance||Instance||A DB2 instance can contain several databases|
|initSID.ora OR SPFILE||DBM CFG andDB CFG||DB2 uses two levels of configuration:- Database Manager Configuration (DBM CFG) (at the instance level)- Database Configuration (DB CFG) (at the database level).Like in Oracle, many of these configuration parameters can be dynamically changed.|
|Table spaces||Table spaces||DB2 supports SMS and DMS Table spaces. DMS Table spaces are similar to Oracle's.|
|Data Blocks||Pages||DB2 supports these page sizes: 4k, 8k, 16k, 32k. A row must fit any of these page sizes. It cannot span to other pages like in Oracle's.|
|Data Files||DMS Table space containers||Containers for DMS table spaces can be raw devices or files.|
|Redo Log Files||Transaction Log Files|
|Data Buffers||Buffer pools.||DB2 does not have a predefined set of bufferpools, but you can create as many as desired. A bufferpool of a given page size must be exist prior to creating a table space with the given page size.|
|SGA||Database Manager and Database Shared memory|
|Library cache||Package cache|
|Large Pool||Utility Heap|
|Data Dictionary Cache||Catalog cache|
|SYSTEM table space||SYSCATSPACE table space|
- The article "DB2 9.7: Run Oracle applications on DB2 9.7 for Linux, UNIX, and Windows" (developerWorks, May 2010) describes the out-of-the-box support for Oracle's SQL and PL/SQL dialects in DB2 9.7.
- "IBM Data Movement Tool" (developerWorks, Nov 2010): Move data from source databases to DB2 the easy way.
- " An inside look at IBM DB2 Advanced Enterprise Servier Edition" (developerWorks, March 2011): Learn about a software bundle that packages DB2 with optimization, development, and management tools.
- The Redbooks® publication Oracle to DB2 Conversion Guide: Compatibility Made Easy is an Oracle to DB2 migration guide. It describes the new DB2 Oracle Database compatibility features, new technology, best practices for moving to DB2, and how to handle some common scenarios.
- The article "Everything you wanted to know about DB2 processes" (developerWorks, Apr 2003) describes the processes that DB2 uses on Linux, UNIX, and Windows, and details their functions.
- The IBM Redbooks publication Oracle to DB2 Conversion Guide for Linux, UNIX, and Windows is a step-by-step guide for migrating from Oracle to DB2.
- Download a free trial version of DB2 for Linux, UNIX, and Windows.
- Now you can use DB2 for free. Download DB2 Express-C, a no-charge version of DB2 Express Edition for the community that offers the same core data features as DB2 Express Edtion and provides a solid base to build and deploy applications.
- Visit the developerWorks resource page for DB2 for Linux, UNIX, and Windows to read articles and tutorials and connect to other resources to expand your DB2 skills.
- Learn about DB2 Express-C, the no-charge version of DB2 Express Edition for the community.