·Î±×½ºÀ§Ä¡ÀÇ ÀûÀýÇÑ °£°ÝÀº ??
ÀÛ¼ºÀÚ °ü¸®ÀÚ ÀÛ¼º½Ã°£ 2006-11-15 21:46:01
 

init<SID>.ora

¿¡º¸¸é

log???_intervalÆĶó¹ÌÅ͸¦ ´Ã¸®½Ã¸é µË´Ï´Ù.



±×¸®°í ´ÙÀ½

SQL> select name,value

2 from v$sysstat

3 where name in('redo buffer allocation retries','redo log space requests',

4 'redo entries');

NAME VALUE

----------------------------------- ------

redo entries 78 -> redo log buffer¸¦ ¾ó¸¸Å­ »ç¿ëÇß³ª



redo buffer allocation retries 0 -> ³»°¡ ¾²·Á´Â °ø°£ÀÌ ¾ÆÁ÷ disk¿¡ ¾²ÀÌÁö ¾Ê¾Æ

±â´Ù¸®´Â Åë°èÁ¤º¸

redo log space requests 0 -> 0ÀÌ ¾Æ´Ï¸é Ȥ½Ã archive log file¿¡ full ÀÖ³ª È®ÀÎ

======================================================================
===========================================



* LGWR°¡ freeing buffer¸¦ Çϴµ¥ ´À·ÁÁö´Â Çö»ó monitor & tuning



- redo log file¿¡ °æÇÕÀÌ ÀÖ´ÂÁö(log files ÀÇ i/o ¿©·¯ disk¿¡¼­ µ¿½Ã¿¡ ¹ß»ýÇϵµ·Ï)

- v$system_event¿¡ log file switch completion¿¡ wait°¡ ÀÖ´ÂÁö.

ÀÖ´Ù¸é redo log file size¸¦ ´Ã¸®ÀÚ

SQL> select event,total_waits,time_waited,average_wait

2 from v$system_event

3 where event like 'log file switch completion%';

- DBWn°¡ ³¡³ªÁö ¾Ê¾Æ Redo log fileÀÇ spiningÀ» À§ÇØ LGWR°¡ waiting

-- redo log groupÀÇ size¿Í ¼ö¸¦ È®ÀÎÇضó...(Áß¿ä)

-- alert.log fileÀ» º¸¸é spining½Ã "CHECKPOINT NOT COMPLETE" °¡ ÀÚÁÖ ¹ß»ý(2~3ºÐ¸¶´Ù)

-- ´ÙÀ½°ú °°ÀÌ v$system_event ¸¦ È®ÀÎÇصµµÊ

SQL> select event,total_waits,time_Waited, average_wait

2 from v$system_event

3 where event like 'log file switch (check%';

-- ´ÙÀ½ parameter È®ÀÎ

SQL> show parameter log_checkpoint



NAME TYPE VALUE

------------------------------------ ------- ------------------------------

log_checkpoint_interval integer 10000

log_checkpoint_timeout integer 1800

log_checkpoints_to_alert boolean FALSE

- archiver°¡ redo log¸¦ archived redo logfile¿¡ ¾²Áö ¾Ê¾Æ LGWR°¡ ±â´Ù¸®´Â°æ¿ì

-- archive device°¡ fullÀÌ ¾Æ´ÑÁö È®ÀÎ

-- redo log group À» Ãß°¡

-- ´ÙÀ½°ú °°ÀÌ v$system_event ¸¦ È®ÀÎ

SQL> select event,total_waits,time_Waited, average_wait

2 from v$system_event

3 where event like 'log file switch (arch';

-- ARCn process°¡ ¹Ù»Ü¶§ LGWR°¡ »õ·Î¿î ARCn process¸¦ ¶ç¿ìµµ·Ï ´ÙÀ½ parameter ¼³Á¤

log_archive_max_processes = n -> n°³ ¸¸Å­ ±îÁö ARCn process¸¦ ¶ç¿ì¸é¼­ ÀÛ¾÷



-- data block, redo log blockÀÌ ±úÁ³´ÂÁö ¾È±úÁ³´ÂÁö È®ÀÎÇÏ´Â mechanismÀÎ

db_block_checksum parameter°¡ true·Î µÇ¾îÀÖ´ÂÁö È®ÀÎ



======================================================================
===========================================



* Cache Hit Ratio : ÀÌ°Í ¿ª½Ã db startup ÀÌÈÄ ¾î´ÀÁ¤µµ ¾ÈÁ¤µÇ°Ô »ç¿ëÇÑ ÈÄ °è»êÇØ¾ß ¸Â´Ù.



SQL> select 1-(phy.value / (cur.value+con.value)) " Cache Hit Ratio"

2 from v$sysstat cur, v$sysstat con, v$sysstat phy

3 where cur.name = 'db block gets'

4 and con.name = 'consistent gets'

5 and phy.name = 'physical reads';



Cache Hit Ratio

----------------

.95751146 > 90%



À§¿¡¼­ db block gets + consistent gets = logical gets À̹ǷÎ

1-(physical gets / logical gets) °¡ cache hit ratioÀÌ´Ù.



* Data³ª application Design¿¡ µû¶ó, data¸¦ accessÇÏ´Â ¹æ¹ý¿¡ µû¶ó ¸¹ÀÌ Â÷ÀÌ°¡ ³ª´Âµ¥

¿¹¸¦ µé¸é full table scan °°Àº °æ¿ì¿¡´Â dirty buffer°¡

MRU endÂÊÀ¸·Î °¡´Â°Ô ¾Æ´Ï¶ó Memory¿¡¼­ »¡¸® ³»¸®·Á´Â ¸ñÀûÀ¸·Î LRU end ÂÊÀ¸·Î °£´Ù.

(ƯÈ÷ OLTP¼º ¾÷¹«¿Í´Â ´Þ¸® DW¼º ¾ø¹«´Â Àå±â°£ full table scanÀ» ÇØ¾ß ÇϹǷΠhit rationº¸´Ù´Â

I/O¿¡ point¸¦ µÎ°í tuningÀ» ÇÑ´Ù.)



DB_BLOCK_BUFFERS¸¦ ´Ã¸®°í ¼Óµµ¸¦ testÇØ º¸¾Æ È¿°ú°¡ ¾øÀ» °æ¿ì¿¡´Â ¹Ýº¹ÀûÀ¸·Î ÇÒ ÇÊ¿ä¾øÀÌ

´äÀÌ ¾Æ´Ï´Ù.


¸ñ·Ï | ÀÔ·Â | ¼öÁ¤ | ´äº¯ | »èÁ¦