Você está na página 1de 1

User Process ™

Oracle Net, CPUs CPUs


tnsnames.ora Operating System

PMON SMON Snn*/Dnn MMAN * MMON * RECO CJQn *


- Cleans User & Server - Crash Recovery Used in Shared Server Dynamically ajusts the The Distributed Job Queue Processes. Mich Talebzadeh
Captures statistics for SQL
Processes Environment Transaction Recovery Selects jobs that need to
- Clean Temp Segments size of SGA components objects which have been mich@peridale.co.uk
- Registers service with Process finds pending be run, .
- Collates Free Space Stack recently modified Version 2, Jan 2006.
listener distributed transactions Replication and Queuing
Space and resolves them NOT PERSISTENT

System Global Area


st *
eque
o nnect R
ing C
Checkpoint Queue DBSNMP *
Incom Database Buffer Cache
Shared Pool Sort
Extent Intelligent Agent
aged blocks (non recurring access) Redo Log
Default Pool
Shared SQL Area Midpoint
Buffer
Library Cache - Hash Value Pool hot area cold area Large Pool
- SQL Source insertion Fixed
Listener Area
User’s privileges - Execution plan
C D D C C D C D UGA * (Static Pnnn *
Overhea
MRU blocks with recurring LRU I/O Buffer d) Parallel Query
access Individual blocks read Area
Data Dictionary Cache read from disk go to

s k
Network Files f (also known as Row Cache)
o the middle UGA Java

loc
n keep recycle nK * 8K *
io go Pool

)B
- listener.ora rs an Response
sc MRPn *

ed
ve e e
l Queue

g
- tnsnames.ora e d d
System tab en
SGA_TARGET (Dynamic) *

(A
im coEnqueues Latches
nt Streams
SGA_MAX_SIZE (Fixed at start-up)

- sqlnet.ora u Parameters hf
ull LRU Request Data Guard

ed
R wit ly to Pool *
s Queue Managed Recovery

ard
ck ate

c
Fork Blo edi
m

Dis
im

DBWn CKPT LGWR ARCn *


Server Process Oracle background processes are persistent Flushes redo log buffer to
processes that make up the instance and they Dirty Buffer threshold - Regularly initiates a Archiver process copies an
online redo log file every 3
will run until the instance is shutdown. They Scan 1/2 LRU list checkpoint. Uses DBWR online redo log file to
seconds, in commits, and
Program Global Area (PGA) Full Checkpoints to write all dirty blocks another location when the
are processes, not programs. There is only checkpoints or when the
one Oracle program on the host, i.e. there is Timeout 3 seconds - Updates data file headers redo log file is filled up or
redo log buffer is 1MB or
UGA only one binary, named oracle. It is just - Shutdown there is a log switch
more than 1/3 full.
Stack executed many times with different names
User Cursor
Space Session State
Data

Oracle supports Bigfile Data


Tablespaces and non- Multiplexed Group 1 Multiplexed Group 2 Multiplexed
Dictionary Multiplexed Data
standard block sizes Data files Data
Control Files Archived Redo
- system01.dbf Data Logs
- sysaux01.dbf Data Files Online Redo Logs
Automatic
Storage Temporary
Management * Tablespace Group Recovery
nk * catalog *
- RBAL Data Files block
- ARBn size RMAN MML
- ASMB

Misc
Oracle managed
Real Application Others * Server Data
Advanced Password Parameter file(s) backups with
Parameter Guard
Clusters * - RVWR, Flashback File file alert_SID.ora Broker RMAN
Queuing * - CTWR, change tracking orapwSID file initSID.ora trace Tape
spfileSID.ora Config
- LMON - ORACM process
- LCKn, - GSD - QMNC
Disk Cross-platform
- LMSn - DIAG - Qnnn transportable To another
- EMNn tablespaces Instance
- LMDn * indicates optional setup

Você também pode gostar