1 Database Architecture

8/9/2019 1 Database Architecture

1/53

Oracle Database Architecture

Prepared ByProduct Alliance Group

Tata Consultancy Services

Ahmedabad


2/53

ArchitecturePhysical Structure

Logical Structure


3/53

ArchitectureInstance

SGA

Redo log

buffer cache

Shared pool

Data Dict.

cache

Library

cache

DBWRSMONPMON CKPTLGWR Others

User

process

Server

processPGA

Controlfiles

Datafiles Archived

log files

Parameter

file

Password

file

Redo logfiles

Database

buffer cache


4/53

Physical StructuresData-files

Control Files

Redo Log Files

Archive Log Files

Parameter Files

Alert and Trace Log Files

Backup Files


5/53

DatafilesContain all the database data.

Data of logical database structures, such as tables

and indexes, is physically stored in the datafilesallocated for a database.

A data-file can be associated with only onedatabase.

Can automatically extend when the database runs

out of space.One or more datafiles form a logical unit of

database storage called a tablespace.


6/53

Control Files Specify the physical structure of the database.

Database name

Names and locations of datafiles and redo log files

Time stamp of database creation Multiplex the control file to protect against a failure

involving the control file.

When an instance of an Oracle database is started, itscontrol file identifies the database and redo log files thatmust be opened for database operation to proceed.

When physical makeup of the database is altered, controlfile is automatically modified by Oracle to reflect thechange.

Used in database recovery.


7/53

Redo Log Files Every Oracle database has a set of two or more redo log

files.

Record all changes made to data.

Multiplexed redo log so that two or more copies of theredo log can be maintained on different disks.

The information in a redo log file is used only to recoverthe database from a system or media failure that preventsdatabase data from being written to the datafiles.

The process of applying the redo log during a recoveryoperation is called rolling forward.


8/53

Parameter FilesParameter files contain a list of configuration

parameters for that instance and database.

Oracle recommends that you create a serverparameter file (SPFILE) as a dynamic means of

maintaining initialization parameters.

A server parameter file lets you store and manage

your initialization parameters persistently in aserver-side disk file.


9/53

Alert and Trace Log Files Each server and background process can write to an

associated trace file.

When an internal error is detected by a process, it dumps

information about the error to its trace file. Some of the information written to a trace file is intended

for the database administrator, while other information isfor Oracle Support Services.

Trace file information is also used to tune applications and

instances. The alert file, or alert log, is a special trace file. The alert

log of a database is a chronological log of messages anderrors.


10/53

Logical StructuresTablespaces

Oracle Data Blocks

Extents

Segments


11/53

Tablespaces

Groups related logical structures together.

One or more datafiles are explicitly created for eachtablespace to physically store the data of all logicalstructures in a tablespace.

Every database contains a SYSTEM and a SYSAUX (indatabase 10g Only).

The default is a smallfile tablespace, which is thetraditional type of Oracle tablespace.

The SYSTEM and SYSAUX tablespaces are created assmallfile tablespaces.

Bigfile tablespaces contain tablespaces made up of singlelarge files rather than numerous smaller ones.

With Bigfile tablespaces database can scale up to 8exabytes in size.


12/53

Oracle Data BlocksAt the finest level of granularity, data is stored in

data blocks.

One data block corresponds to a specific numberof bytes of physical database space on disk.

The standard block size is specified by theDB_BLOCK_SIZE initialization parameter.

In addition, you can specify up to five other block

sizes. A database uses and allocates free database space

in Oracle data blocks.


13/53

Extents

The next level of logical database space is

an extent.

An extent is a specific number of

contiguous data blocks, obtained in a single

allocation, used to store a specific type of

information.


14/53

Segments

Above extents, the level of logical database

storage is a segment.

A segment is a set of extents allocated for a certainlogical structure. The following table describes the

different types of segments.

Data segment

Index segment

Temporary segment

Rollback segment


15/53

Graphically


16/53

Database

Logical Physical

Tablespace Data file

OS blockOracle

block

Segment

Extent

Database Storage Hierarchy


17/53

Overview of Schemas and Common

Schema Objects

Tables

Indexes

Views

Clusters

Synonyms


18/53

Tables

Tables are the basic unit of data storage in

an Oracle database.

Each table has column and row.A table that has an employee database, for

example, can have a column called

employee number, and each row in thatcolumn is an employee's number.


19/53

Indexes

Optional structures associated with tables.

Can be created to increase the performance of dataretrieval.

When processing a request, Oracle can use some or all ofthe available indexes to locate the requested rowsefficiently.

Useful when applications frequently query a table for arange of rows

Indexes are created on one or more columns of a table.

After it is created, it is automatically maintained and usedby Oracle.

Changes to table data (such as adding new rows, updatingrows, or deleting rows) are automatically incorporated intoall relevant indexes.


20/53

Views Customized presentations of data in one or more tables or

other views.

Can also be considered a stored query.

Do not actually contain data.

They derive their data from the base tables on which theyare based.

Can be queried, updated, inserted into, and deleted from,with some restrictions.

All operations performed on a view actually affect the basetables of the view.

Provide an additional level of table security by restrictingaccess to a predetermined set of rows and columns of atable.

Hide data complexity and store complex queries.


21/53

ClustersClusters are groups of one or more tables

physically stored together because they sharecommon columns and are often used together.

Because related rows are physically storedtogether, disk access time improves.

Do not affect application design.

Whether a table is part of a cluster is transparent to

users and to applications.Data stored in a clustered table is accessed by

SQL in the same way as data stored in a nonclustered table.


22/53

Synonyms

A synonym is an alias for any table, view,

materialized view, sequence, procedure,

function, package, type, Java class schemaobject, user-defined object type, or another

synonym.

Because a synonym is simply an alias, it

requires no storage other than its definition

in the data dictionary.


23/53

Overview of the Oracle Data

Dictionary It is a set of tables and views that are used as a

read-only reference about the database.

Stores information about both the logical andphysical structure of the database.

A data dictionary also stores

The valid users of an Oracle database

Information about integrity constraints definedfor tables in the database

The amount of space allocated for a schemaobject and how much of it is in use


24/53

Overview of the Oracle Data

Dictionary

Automatically updated by Oracle when the

structure of the database is altered.

The database relies on the data dictionary to

record, verify, and conduct ongoing work.


25/53

Overview of the Oracle InstanceThe combination of

Memory buffers

Background processes


26/53

Instance Memory StructuresSystem Global Area

Database Buffer Cache

Redo Log Buffer

Shared Pool

Large pool

Java pool

Program Global Area


27/53

System Global Area

Instance

SGA

Redo log

buffer cache

Shared pool

Data Dict.

cache

Library

cache Database

buffer cache

SGA is a shared memory region that contains data andcontrol information for one Oracle instance.

Oracle allocates the SGA when an instance starts and de-

allocates it when the instance shuts down.

Each instance has its own SGA

Users connected to database share the data in the SGA.


28/53

System Global Area

Database Buffer Cache

The database buffer cache stores copies of datablocks that have been retrieved from the data files.

It enables great performance gains when youobtain and update data.

It is managed through a least recently used (LRU)algorithm.

DB_CACHE_SIZEdetermines thesize of database buffer cache Database buffer

cache


29/53

System Global Area

Redo Log Buffer

The redo log buffer cache records all changes made to thedatabase data blocks.

Its primary purpose is recovery. Changes recorded within are called redo entries.

Redo entries contain information to reconstruct or redochanges.

Size is defined by LOG_BUFFER.

Redo log

buffer cache


30/53

System Global Area

Shared Pool Used to store the most recently executed SQL statements

and the most recently used data definitions.

It consists of two key performance-related memorystructures

Library cache

Data dictionary cache

Sized by the parameter SHARED_POOL_SIZE

ALTER SYSTEM SET SHARED_POOL_SIZE = 64M;

Shared pool

Data dictionary

cache

Library

cache


31/53

System Global AreaLibrary cache The library cache stores information about the most

recently used SQL and PL/SQL statements.

The library cache enables the sharing of commonly usedstatements

Is managed by a least recently used (LRU) algorithm

Consists of two structures:

Shared SQL area

Shared PL/SQL area Has its size determined by the shared pool sizing


32/53

System Global AreaData dictionary cache

The data dictionary cache is a collection of the mostrecently used definitions in the database.

It includes information about database files, tables,indexes, columns, users, privileges, and other databaseobjects.

During the parse phase, the server process looks at the datadictionary for information to resolve object names and

validate access. Caching the data dictionary information into memory

improves response time on queries.

Size is determined by the shared pool sizing.


33/53

System Global AreaLarge pool

The large pool is an optional area of memory in the SGAconfigured only in a shared server environment.

It relieves the burden placed on the shared pool.

This configured memory area is used for session memory(UGA), I/O slaves, and backup and restore operations.

Unlike the shared pool, the large pool does not use an LRU

list. Sized by LARGE_POOL_SIZE.

ALTER SYSTEM SET LARGE_POOL_SIZE = 64M;


34/53

System Global AreaJava pool

The Java pool services the parsing

requirements for Java commands.Required if installing and using Java.

It is stored much the same way as PL/SQL indatabase tables.

It is sized by the JAVA_POOL_SIZEparameter.


35/53

Program Global Area (PGA)

The PGA is memory reserved for each user process that

connects to an Oracle database.

User

process

Stackspace

Sessioninformation

sort area, cursorinformation

SGA

Shared SQL areas

SGASession

information

PGA Dedicated server Shared server

Serverprocess

Shared SQL areas

Stackspace

sort area, cursorinformation


36/53

User Process A user process is a program that requests interaction with

the Oracle server.

It must first establish a connection.

It does not interact directly with the Oracle server.

Database user

Serverprocess

Userprocess

Connection

established


37/53

Server Process

A server process is a program that directly interacts with

the Oracle server. It fulfills calls generated and returns results.

Can be dedicated or shared server.

Connection

establishedSession created

Database user

Userprocess

Oracle server

Serverprocess


38/53

Oracle Background Processes The relationship between the physical and memory

structures is maintained and enforced by Oraclesbackground processes.

Mandatory background processesDBWn PMON CKPT

LGWR SMON RECO

Optional background processes

ARCn LMON Snnn

QMNn LMDn CJQ0 Pnnn

LCKn Dnnn


39/53

Instance

SGA

Database

buffercache

DBWn

Controlfiles

Datafiles

Redo logfiles

Database

DBWn writes when:

Checkpoint

Dirty buffers threshold reached

No free buffers

Timeout

RAC ping request

Tablespace offline

Tablespace read only

Table DROP orTRUNCATE

Tablespace BEGIN BACKUP

Database Writer (DBWn)


40/53

Instance

SGA

DBWn

Redo log

buffer

ControlfilesDatafiles Redo logfiles

LGWR

Database

LGWR writes:

At commit

When one-third full

When there is 1 MB ofredo

Every 3 seconds

Before DBWn writes

Log Writer (LGWR)


41/53

Instance

SGA

SMON

Control

files

Data

files

Redo log

files

Instance

SGA

SMON

Database

Responsibilities:

Instance recovery:

Rolls forward changes in

the redo logs

Opens the database for

user access Rolls back uncommitted

transactions

Coalesces free space ever 3 sec

Deallocates temporary

segments

System Monitor (SMON)


42/53

Process Monitor (PMON)

Cleans up after

failed processes by:

Rolling back the transaction

Releasing locks

Releasing other resources

Restarts dead dispatchers

Instance

SGA

PMON

PGA area


43/53

Instance

SGA

DWW0

Redo Log

Buffer

LGWR

Instance

SGA

DBWn LGWR CKPT

Controlfiles

Datafiles

Redo logfiles

Checkpoint (CKPT)

Responsible for:

Signaling DBWn at

checkpoints

Updating datafileheaders with

checkpoint

information

Updating control files

with checkpoint

information


44/53

Archiver (ARCn) Optional background process

Automatically archives online redo logs when

ARCHIVELOG mode is set Preserves the record of all changes made to the database

Controlfiles

Datafiles

Redo logfiles

ArchivedRedo log

files

ARCnARCn


45/53

Starting Up a Database

OPEN

MOUNT

NOMOUNT

SHUTDOWN

All files opened as

described by the control

file for this instance

Control file

opened for this

instance

STARTUP

SHUTDOWN

Instance

started pfile

or spfile used


46/53

Shutting Down the Database

Shutdown Mode:

NORMAL

TRANSACTIONAL

IMMEDIATE

ABORT

A

x

x

x

x

T

x

x

o

o

I

x

x

x

o

Shutdown Mode

Allow new connections

Wait until current sessions end

Wait until current transactions end

Force a checkpoint and close files

N

x

o

o

o

YESNOx

o


47/53

Shutdown OptionsDuring a

Shutdown Normal,

Shutdown Transactional

or

Shutdown Immediate

Consistent Database

(clean database)

On the way down:

Database buffer

cache written to

the data files Uncommitted

changes rolled

back

Resources

released.

On the way up:

No instance

recovery


48/53

Shutdown OptionsDuring a

Shutdown Abort

or

Instance Failure

or

Startup Force

Inconsistent Database

(dirty database)

On the way down:

Modified buffers

are not written to

the data files

Uncommitted

changes are not

rolled back

On the way up:

Redo logs used to

reapply changes

Undo segments

used to roll back

uncommitted

changes

Resources

released


49/53

How Oracle Works

1. An instance has started on the computer running Oracle

2. A computer running an application (a local computer orclient workstation) runs the application in a user process.

The client application attempts to establish a connection tothe server using the proper Oracle Net Services driver.

3. The server is running the proper Oracle Net Servicesdriver. The server detects the connection request from theapplication and creates a dedicated server process on

behalf of the user process.

4. The user runs a SQL statement and commits thetransaction.


50/53

How Oracle Works

5. The server process receives the statement and checks theshared pool for any shared SQL area that contains asimilar SQL statement. If a shared SQL area is found, thenthe server process checks the user's access privileges to therequested data, and the previously existing shared SQL

area is used to process the statement. If not, then a newshared SQL area is allocated for the statement, so it can be

parsed and processed.

6. The server process retrieves any necessary data valuesfrom the actual datafile (table) or those stored in the SGA.

7. The server process modifies data in the system global area.The DBWnprocess writes modified blocks permanently todisk when doing so is efficient. Because the transaction iscommitted, the LGWR process immediately records thetransaction in the redo log file.


51/53

How Oracle Works

8. If the transaction is successful, then the serverprocess sends a message across the network tothe application. If it is not successful, then anerror message is transmitted.

9. Throughout this entire procedure, the otherbackground processes run, watching forconditions that require intervention. In addition,the database server manages other users'transactions and prevents contention betweentransactions that request the same data.


52/53

Q & A


53/53

Thank You

1 Database Architecture

Documents