You are on page 1of 1

User Process

Operating System
SMON
Snn*/Dnn
Used in Shared Server Environment Dynamically ajusts the size of SGA components Captures statistics for SQL objects which have been recently modified CPUs

Oracle Net, tnsnames.ora

CPUs

PMON MMON *
The Distributed Transaction Recovery Process finds pending distributed transactions and resolves them Job Queue Processes. Selects jobs that need to be run, . Replication and Queuing NOT PERSISTENT

MMAN *

RECO

CJQn *

- Cleans User & Server Processes - Registers service with listener

- Crash Recovery - Clean Temp Segments - Collates Free Space

Stack Space

Mich Talebzadeh mich@peridale.co.uk Version 2, Jan 2006.

ing Incom

st * eque ect R n n o C

System Global Area


Database Buffer Cache
aged blocks (non recurring access) hot area D C D C D D C Midpoint insertion cold area Checkpoint Queue Sort Extent Pool

DBSNMP *
Intelligent Agent

Shared Pool Shared SQL Area


- Hash Value - SQL Source - Execution plan C

Listener
MRU
read
s

Library Cache

Default Pool Large Pool UGA * LRU


I/O Buffer Area UGA Response Queue Request Queue

Redo Log Buffer

Users privileges

Pnnn *
Parallel Query

Fixed Area (Static Overhea d)

Data Dictionary Cache f (also known as Row Cache)


(A g ed )B loc k

go an sc e l d tab en ull LRU hf wit ly to s ck ate Blo edi m im Dis c ard ed

Network Files - listener.ora - tnsnames.ora - sqlnet.ora


keep Latches 8K * System Parameters

Streams Pool *

SGA_MAX_SIZE (Fixed at start-up)


Java Pool

o n io rs ve e e d im coEnqueues nt

SGA_TARGET (Dynamic) *

blocks with recurring access Individual blocks read from disk go to the middle nK * recycle

MRPn *
Data Guard Managed Recovery

Fork

Server Process
Dirty Buffer threshold Scan 1/2 LRU list Full Checkpoints Timeout 3 seconds - Shutdown

DBWn

CKPT
- Regularly initiates a checkpoint. Uses DBWR to write all dirty blocks - Updates data file headers

LGWR

ARCn *
Archiver process copies an online redo log file to another location when the redo log file is filled up or there is a log switch

Program Global Area (PGA)

UGA

Stack Space

User Session Data

Cursor State

Oracle background processes are persistent processes that make up the instance and they will run until the instance is shutdown. They are processes, not programs. There is only one Oracle program on the host, i.e. there is only one binary, named oracle. It is just executed many times with different names

Flushes redo log buffer to online redo log file every 3 seconds, in commits, and checkpoints or when the redo log buffer is 1MB or more than 1/3 full.

Oracle supports Bigfile Tablespaces and nonstandard block sizes

Data Dictionary Data files - system01.dbf

Multiplexed Control Files Data Files


Data

Multiplexed Group 1 Data

Multiplexed Group 2 Data

Automatic Storage Management *


nk * block size

- sysaux01.dbf

Online Redo Logs


Recovery catalog *

Multiplexed Archived Redo Logs

- RBAL - ARBn - ASMB

Temporary Tablespace Group Data Files

RMAN
Oracle managed backups with RMAN

MML

Real Application Clusters *

Advanced Queuing *

Others *
- RVWR, Flashback - CTWR, change tracking
process

Password File orapwSID

Server Parameter file spfileSID.ora

Parameter file initSID.ora

Misc file(s) alert_SID.ora trace

Data Guard Broker Config Disk

- LMON - LCKn, - LMSn - LMDn

- ORACM - GSD - DIAG

Tape

- QMNC - Qnnn - EMNn

* indicates optional setup

Cross-platform transportable tablespaces

To another Instance

You might also like