SlideShare une entreprise Scribd logo
1  sur  53
Oracle Wait Events
That Everyone Should Know
Kerry Osborne
Senior Oracle Guy
What Are Wait Events?
 Basically a section of code
 Uses gettimeofday
 All time is stuck in a bucket with a name
 Multiple things can be stuck in the same bucket
 Oracle is very well instrumented (but It’s not 100%)
 There are 41 classes of wait events in 10.2.0.3
 There are 878 wait events in 10.2.0.3
 209 enqueue events
 29 latch events
 41 I/O events
What do you do with Wait Events?
 See where Oracle is spending it’s time
 Create a Resource Profile
 Aggregate the time by Wait Event
 Order the profile by time
 Include # of occurrences and avg. time/event
 Shows where to focus
 Shows how much improvement can be made
Profiles
 Basic Tool – maybe the most important
 Shows up in Statspack / AWR reports
 Shows up in tkprof output (10g)
 Can be pulled out of trace files
 Can be pulled from V$ tables (v$system_event)
 Can be pulled from ash tables
Oracle is very well instrumented
Profiles
Top 5 Timed Events
~~~~~~~~~~~~~~~~~~ % Total
Event Waits Time (s) Ela Time
-------------------------------------------- ------------ ----------- --------
CPU time 37,297 52.27
SQL*Net message from dblink 7,065,802 11,312 15.85
async disk IO 943,852 5,265 7.38
db file sequential read 13,042,432 3,493 4.90
direct path write 108,862 3,410 4.78
-------------------------------------------------------------
Why should developers know this stuff?
1. Because you need to know
where your code is spending it’s
time
2. Scalability is Important
 1. These are the events you should know by heart
 2. The other 260 or so you can figure out
 3. You need to know what causes them
 4. You need to know what values are reasonable
 5. You need to know what you can do to fix them
Top 10 List of Wait Events
select name, parameter1, parameter2, parameter3
from v$event_name
where name like nvl('&event_name',name)
order by name
 Not a wait event – but often included in a profile
 Time spent but not accounted for in other buckets
 Primarily time spent doing lio (we hope)
 We want it to be on top
 It may be an indicator of inefficient plans
CPU
 Single block read
 Usually index block or data block via rowid
 Can be done by fts to get chained rows
 P1=file, P2=block, P3=# of blocks (always 1)
 Always a user’s server process reading into buffer cache
 Reasonable e values: <10ms
DB File Sequential Read
WAIT #14: nam='db file sequential read' ela= 36745 file#=1 block#=67745 blocks=1
obj#=84920
 Do Less Work
 tune SQL
 reduce lio’s 7 / obj rule of thumb
 Explain plan lies (see notes on this slide)
 bigger buffer cache
 Do the Work Faster
 Speed up I/O
 Call Jack
DB File Sequential Read – “fixes”
 Explain plan appears to executes a separate code path
 Even if it didn’t there is no guarantee TEST matches PROD
 Best bet is to execute and look at V$SQL_PLAN
 This is easy in 10g with dbms_xplan
select * from table(dbms_xplan.display_cursor('&sql_id','&child_no',''))
 See an example in the notes section for this slide
Digression – Explain Plan Lies
 Multi block read
 Usually full table scan or index fast full scan
 P1=file, P2=block, P3=# of blocks (always < MBRC)
 Always a user’s server process reading into buffer cache
 Reasonable ela values: <10ms
 Relevant parameters:
 DBFMBRC, System Stats: MBRC,
 _db_file_exec_read_count, _db_file_optimizer_read_count
DB File Scattered Read
WAIT #14: nam='db file scattered read' ela= 19389 file#=139 block#=44 blocks=5
obj#=83580
 Do Less Work (fewer multi-block reads)
 MBRC
 Better Indexes
 System Stats
 bigger buffer cache
 Tune SQL
 Do the Work Faster
 Speed up I/O
DB File Scattered Read – “fixes”
 Usually sorting to Temp
 Can also be parallel query
 Could also be insert append, etc…
 Reasonable ela values: <20ms
 Relevant parameters:
 DBFMBRC
 _db_file_direct_io_count (1M)
Direct Path Read/Write
WAIT #10: nam='direct path write' ela= 4475 p1=401 p2=1518353 p3=57
WAIT #10: nam='direct path read' ela= 16770 p1=401 p2=1482031 p3=63
 Adjust PGA_AGGREGATE_TARGET
 Turn off PX query
 Call Randy
Direct Path Read/Write – “fixes”
 User process wait for LGWR to flush dirty buffers on commit
 Synchronous write event
 P1=log buffer block
 Reasonable ela values: <4ms
Log File Sync
WAIT #16: nam='log file sync' ela= 1286 buffer#=11910 p2=0 p3=0 obj#=84920
 Do Less Work (fewer commits)
 Autocommit?
 Row at a time processing with commits?
 Do the Work Faster
 Speed Up I/O
 No RAID 5
 No Contention (other db, arch, ASM)
 SS Disk
 Improve LGWR Scheduling
 Renice
Log File Sync – “fixes”
 Wait for LGWR to write to current log files
 Not necessarily synchronous write event
 System I/O event
 Reasonable ela values: <4ms
 Occurs as follows:
 Every 3 seconds
 Log_buffer is 1/3 full or redo = 1M ( default _log_io_size)
 Commit or rollback by user session
 RAC needs to transfer a dirty block
Log File Parallel Write
 Often fixed by fixing Log File Sync (i.e. speed up i/o, etc)
 Force LGWR to write more often
Log File Parallel Write – “fixes”
 Happens when database can’t switch to next log file
 Freezes the whole database
 Several Flavors
 Checkpoint incomplete
 Archiving needed
 Reasonable ela values: 0ms
Log file switch …
WAIT #9: nam='log file switch (checkpoint incomplete)' ela= 986212 p1=0 p2=0 p3=0
 Do less work
 Make total online log space bigger
 Force more frequent incremental checkpoints
 Speed up i/o
Log file switch … – “fixes”
 Contention event for the same block
 Read by other session (new event in 10g)
 Held by other session in incompatible mode
 New event in 10g – read by other session
 Multiple sections of code throw time in this bucket
 P1=file, P2=block, P3=reason code
 See metalink note (34405.1) for details
Buffer Busy Waits
WAIT #7: nam='read by other session' ela= 3251 file#=4 block#=188557 class#=1
obj#=53707 tim=1183096831514887
 It’s complicated – but basically - eliminate the contention
 For Read by Other Session
 Do Less Work (tune SQL to do less lio)
 Eliminate i/o – bigger buffer cache, etc…
 Speed up i/o
 For Others
 Find type of hot objects and statements suffering
 Address issue - ASSM, Freelist groups (RAC), etc..
Buffer Busy Waits - “fixes”
 No place to put a new block
Free Buffer Waits
 Do Less Work
 Reduce the amount of lio
 Make more buffer space available
 Add memory to the buffer cache
 Speed up DBWR to flush blocks more quickly
 Async_io
 More dbwr processes
 Renice
Free Buffer Waits - “fixes”
 Database is idle – waiting on next request from client
 Often marked as an idle event which it may be
 Time distribution can be skewed
 Common technique to ignore anything over 1 sec
 Reasonable e values: <2ms
 Protocol should be tcp (or beq for local processes)
SQL*Net message from client
WAIT #1: nam='SQL*Net message from client' ela= 336 driver id=1650815232 #bytes=1 p3=0 obj#=83660
 Speed up app server
 Reduce the number of calls
 Use the right protocol (tcp to beq)
 Call network guy
 Get the users to type faster
SQL*Net message from client – “fixes”
 More than 1 packet required to send SQL statement
SQL*Net more data from client
 Don’t send 50K SQL statements
 Use packages
SQL*Net more data from client – “fixes”
 Time to send a SQL*Net message to a client (sort of)
 Actually time it took to write the return data from Oracle’s userland
SDU buffer to OS kernel-land TCP socket buffer (Tanel Poder)
 Often marked as an idle event which it may be
 Reasonable e values: <1ms (micro seconds actually)
 P1=bytes (usually 1 ???)
SQL*Net message to Client
WAIT #1: nam='SQL*Net message to client' ela= 2 driver id=1650815232 #bytes=1 p3=0 obj#=83660
 Use correct protocol
 Fix network issue
 Call Andy
SQL*Net message to Client – “fixes”
 When more than one packet required
 Rows spanning block
 LOB’s
 Array fetches
 Often marked as an idle event which it may be
 Reasonable e values: <1ms
 P1=bytes (usually 1 ???)
SQL*Net more data to client
WAIT #1: nam='SQL*Net message to client' ela= 2 driver id=1650815232 #bytes=1 p3=0 obj#=83660
 Fix chained/migrated rows
 Don’t use LOBs
 Nothing to fix
SQL*Net more data to client – “fixes”
 Locks for objects
 Que up in order
 9i lumped them all together
 10g has 209 separate events
Enqueues
 P1 has encoded type and mode
Enqueue in 9i
SELECT
chr(bitand(p1,-16777216)/16777215)||
chr(bitand(p1, 16711680)/65535) Type,
mod(p1,16) lmode
from v$session_wait
where event=‘enqueue’;
 Several Major Categories
 TM – table modification
 TX – Transaction locks
 UL – user lock
 CI – Cross Instance
 CU – Cursor Bind
 HW – High Water
 RO – Reusable Object
 ST – Space Transaction
 TS – Temporary Space
 Parameters depend on type
Enqueue in 10g
 User can’t modify block until other user commits
Enqueue: TX row lock
 Use ash tables to find statements waiting
 Fix the code
 Or do what one of our clients did:
write some code to kill any processes blocking for more
than 60 seconds – lot’s easier than changing the app
Enqueue: TX row lock – “fixes”
 Waiting on exclusive access to a sequence
 Seems minor but can actually be a big issue on some
systems
Enqueue: SQ
 Cache the sequence
 Use NOORDER in RAC
Enqueue: SQ – “fixes”
 Locks internal memory structures
 Not ordered
 Very light weight
 Kid Asking for Candy Model
 Spins vs. Sleeps (_spin_count, _latch_class_X)
 Wait events do not include time spent spinning
 9i lumped them all together
 10g has 28 separate events
Latches
 All dumped in the Latch Free event
 P2 has latch#
Latches in 9i
WAIT #9: nam='latch free' ela= 185 p1=-4611686003650090584 p2=157 p3=0
SQL> Select name from v$latch where latch#=157;
NAME
----------------------------------------
library cache
 Primarily Deal with major SGA areas
 Shared Pool
 Latch: shared pool
 Latch: library cache
 Latch row cache
 Buffer Cache
 Latch: cache buffer chains
 Latch: cache buffers lru chain
 Log Buffer
 Latch: redo …
 Parameters depend on type (see v$event_name)
Latches in 10g
 Each block hashes to a chain
 Several chains per latch
 Several latches per instance (default based on cache size)
Latch: cache buffers chains
 Find hot blocks
 Fix bad SQL
 Maybe Increase number of latches
 Maybe decrease rows per block
Latch: cache buffers chains –”fixes”
 Shared Pool Contention
 Due primarily to parsing
 See also latch: library cache …
Latch: shared pool
WAIT #28: nam='latch: shared pool' ela= 751 address=1611562656 number=213 tries=1 j#=46024
 Use Bind Variables
 CURSOR_SHARING = FORCE
 Increase SHARED_POOL
– this is where Darcy came in and sat in my office so
I didn’t get much more done –
You can thank Darcy later!
Latch: shared pool – “fixes”
 Some tools show this (Hotsos Profiler)
 Indicates lost time due to rounding (small)
 Or lost time due to CPU contention (large)
Unaccounted for
Tracing
 alter session set events '10046 trace name context
forever, level 8';
 Spits out wait events, plans (real), stats
 Level 12 includes bind variables – makes file very big
 Puts file in user_dump_dest (ls –altr)
 Can be executed from logon trigger, in-line, etc…
 Scoping is important (i.e. when it’s turned on and off)
 Note: bug in early versions of 9.2, fixed in 9.2.0.6
Trace File Contents
 WAIT
 FETCH
 EXEC
 PARSE
 STAT
 XCTEND
 PARSING IN CURSOR
Raw Trace File
 /opt/oracle/admin/LAB102/udump/lab102_ora_4207.trc
 Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Production
 ORACLE_HOME = /opt/oracle/product/db/10.2.0/db1
 System name: Linux
 Node name: homer
 Release: 2.6.9-34.ELhugemem
 Version: #1 SMP Fri Feb 24 17:04:34 EST 2006
 Machine: i686
 Instance name: LAB102
 Redo thread mounted by this instance: 1
 Oracle process number: 20
 Unix process pid: 4207, image: oracle@homer (TNS V1-V3)
 *** ACTION NAME:() 2007-08-16 13:48:14.571
 *** MODULE NAME:(SQL*Plus) 2007-08-16 13:48:14.571
 *** SERVICE NAME:(SYS$USERS) 2007-08-16 13:48:14.571
 *** SESSION ID:(143.189) 2007-08-16 13:48:14.571
 =====================
 PARSING IN CURSOR #7 len=68 dep=0 uid=61 oct=42 lid=61 tim=1159462982979740 hv=740818757 ad='30663e48'
 alter session set events '10046 trace name context forever, level 8'
 END OF STMT
 EXEC #7:c=0,e=269,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=1,tim=1159462982979717
 WAIT #7: nam='SQL*Net message to client' ela= 7 driver id=1650815232 #bytes=1 p3=0 obj#=-1 tim=1159462982980778
 WAIT #7: nam='SQL*Net message from client' ela= 119 driver id=1650815232 #bytes=1 p3=0 obj#=-1 tim=1159462982981008
 =====================
 PARSING IN CURSOR #8 len=44 dep=0 uid=61 oct=3 lid=61 tim=1159463023994427 hv=761757617 ad='54738434'
 select avg(col1) from skew
 where rownum < 10
 END OF STMT
 PARSE #8:c=4000,e=3904,p=0,cr=0,cu=0,mis=1,r=0,dep=0,og=1,tim=1159463023994411
 EXEC #8:c=0,e=185,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=1,tim=1159463023994844
 WAIT #8: nam='SQL*Net message to client' ela= 9 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023994980
 WAIT #8: nam='db file scattered read' ela= 218 file#=4 block#=21900 blocks=5 obj#=53707 tim=1159463023995544
 FETCH #8:c=0,e=665,p=5,cr=4,cu=0,mis=0,r=1,dep=0,og=1,tim=1159463023995732
 WAIT #8: nam='SQL*Net message from client' ela= 157 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996048
 FETCH #8:c=0,e=7,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=0,tim=1159463023996183
 WAIT #8: nam='SQL*Net message to client' ela= 7 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996312
 WAIT #8: nam='SQL*Net message from client' ela= 298 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996669
 XCTEND rlbk=0, rd_only=1
 STAT #8 id=1 cnt=1 pid=0 pos=1 obj=0 op='SORT AGGREGATE (cr=4 pr=5 pw=0 time=736 us)'
 STAT #8 id=2 cnt=9 pid=1 pos=1 obj=0 op='COUNT STOPKEY (cr=4 pr=5 pw=0 time=846 us)'
 STAT #8 id=3 cnt=9 pid=2 pos=1 obj=53707 op='TABLE ACCESS FULL SKEW (cr=4 pr=5 pw=0 time=639 us)'
 WAIT #0: nam='log file sync' ela= 680 buffer#=4862 p2=0 p3=0 obj#=53707 tim=1159463039852003
Questions?
Kerry.Osborne@enkitec.com

Contenu connexe

Similaire à Oracle Wait Events That Everyone Should Know.ppt

SQL Server It Just Runs Faster
SQL Server It Just Runs FasterSQL Server It Just Runs Faster
SQL Server It Just Runs FasterBob Ward
 
Wait events
Wait eventsWait events
Wait eventsAnu Rana
 
Oracle database performance tuning
Oracle database performance tuningOracle database performance tuning
Oracle database performance tuningAbishek V S
 
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert Bialek
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert BialekTrivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert Bialek
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert BialekTrivadis
 
Adventures in Dataguard
Adventures in DataguardAdventures in Dataguard
Adventures in DataguardJason Arneil
 
Treasure Data and AWS - Developers.io 2015
Treasure Data and AWS - Developers.io 2015Treasure Data and AWS - Developers.io 2015
Treasure Data and AWS - Developers.io 2015N Masahiro
 
Clug 2011 March web server optimisation
Clug 2011 March  web server optimisationClug 2011 March  web server optimisation
Clug 2011 March web server optimisationgrooverdan
 
Scaling Apache Storm - Strata + Hadoop World 2014
Scaling Apache Storm - Strata + Hadoop World 2014Scaling Apache Storm - Strata + Hadoop World 2014
Scaling Apache Storm - Strata + Hadoop World 2014P. Taylor Goetz
 
Oracle Client Failover - Under The Hood
Oracle Client Failover - Under The HoodOracle Client Failover - Under The Hood
Oracle Client Failover - Under The HoodLudovico Caldara
 
GC free coding in @Java presented @Geecon
GC free coding in @Java presented @GeeconGC free coding in @Java presented @Geecon
GC free coding in @Java presented @GeeconPeter Lawrey
 
Data Grids with Oracle Coherence
Data Grids with Oracle CoherenceData Grids with Oracle Coherence
Data Grids with Oracle CoherenceBen Stopford
 
Lotusphere 2007 AD505 DevBlast 30 LotusScript Tips
Lotusphere 2007 AD505 DevBlast 30 LotusScript TipsLotusphere 2007 AD505 DevBlast 30 LotusScript Tips
Lotusphere 2007 AD505 DevBlast 30 LotusScript TipsBill Buchan
 
Network and distributed systems
Network and distributed systemsNetwork and distributed systems
Network and distributed systemsSri Prasanna
 
Solve the colocation conundrum: Performance and density at scale with Kubernetes
Solve the colocation conundrum: Performance and density at scale with KubernetesSolve the colocation conundrum: Performance and density at scale with Kubernetes
Solve the colocation conundrum: Performance and density at scale with KubernetesNiklas Quarfot Nielsen
 
AF Ceph: Ceph Performance Analysis and Improvement on Flash
AF Ceph: Ceph Performance Analysis and Improvement on FlashAF Ceph: Ceph Performance Analysis and Improvement on Flash
AF Ceph: Ceph Performance Analysis and Improvement on FlashCeph Community
 
Clug 2012 March web server optimisation
Clug 2012 March   web server optimisationClug 2012 March   web server optimisation
Clug 2012 March web server optimisationgrooverdan
 

Similaire à Oracle Wait Events That Everyone Should Know.ppt (20)

SQL Server It Just Runs Faster
SQL Server It Just Runs FasterSQL Server It Just Runs Faster
SQL Server It Just Runs Faster
 
Wait events
Wait eventsWait events
Wait events
 
Oracle database performance tuning
Oracle database performance tuningOracle database performance tuning
Oracle database performance tuning
 
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert Bialek
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert BialekTrivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert Bialek
Trivadis TechEvent 2016 Oracle Client Failover - Under the Hood by Robert Bialek
 
Adventures in Dataguard
Adventures in DataguardAdventures in Dataguard
Adventures in Dataguard
 
Treasure Data and AWS - Developers.io 2015
Treasure Data and AWS - Developers.io 2015Treasure Data and AWS - Developers.io 2015
Treasure Data and AWS - Developers.io 2015
 
Clug 2011 March web server optimisation
Clug 2011 March  web server optimisationClug 2011 March  web server optimisation
Clug 2011 March web server optimisation
 
Wait events
Wait eventsWait events
Wait events
 
Scaling Apache Storm - Strata + Hadoop World 2014
Scaling Apache Storm - Strata + Hadoop World 2014Scaling Apache Storm - Strata + Hadoop World 2014
Scaling Apache Storm - Strata + Hadoop World 2014
 
Oracle Client Failover - Under The Hood
Oracle Client Failover - Under The HoodOracle Client Failover - Under The Hood
Oracle Client Failover - Under The Hood
 
GC free coding in @Java presented @Geecon
GC free coding in @Java presented @GeeconGC free coding in @Java presented @Geecon
GC free coding in @Java presented @Geecon
 
Data Grids with Oracle Coherence
Data Grids with Oracle CoherenceData Grids with Oracle Coherence
Data Grids with Oracle Coherence
 
Lotusphere 2007 AD505 DevBlast 30 LotusScript Tips
Lotusphere 2007 AD505 DevBlast 30 LotusScript TipsLotusphere 2007 AD505 DevBlast 30 LotusScript Tips
Lotusphere 2007 AD505 DevBlast 30 LotusScript Tips
 
Network and distributed systems
Network and distributed systemsNetwork and distributed systems
Network and distributed systems
 
Solve the colocation conundrum: Performance and density at scale with Kubernetes
Solve the colocation conundrum: Performance and density at scale with KubernetesSolve the colocation conundrum: Performance and density at scale with Kubernetes
Solve the colocation conundrum: Performance and density at scale with Kubernetes
 
AF Ceph: Ceph Performance Analysis and Improvement on Flash
AF Ceph: Ceph Performance Analysis and Improvement on FlashAF Ceph: Ceph Performance Analysis and Improvement on Flash
AF Ceph: Ceph Performance Analysis and Improvement on Flash
 
Clug 2012 March web server optimisation
Clug 2012 March   web server optimisationClug 2012 March   web server optimisation
Clug 2012 March web server optimisation
 
Data race
Data raceData race
Data race
 
AWR Sample Report
AWR Sample ReportAWR Sample Report
AWR Sample Report
 
ASH and AWR on DB12c
ASH and AWR on DB12cASH and AWR on DB12c
ASH and AWR on DB12c
 

Plus de TricantinoLopezPerez

Plus de TricantinoLopezPerez (11)

con9578-2088758.pdf
con9578-2088758.pdfcon9578-2088758.pdf
con9578-2088758.pdf
 
linux-memory-explained.pdf
linux-memory-explained.pdflinux-memory-explained.pdf
linux-memory-explained.pdf
 
sqltuning101-170419021007-2.pdf
sqltuning101-170419021007-2.pdfsqltuning101-170419021007-2.pdf
sqltuning101-170419021007-2.pdf
 
pdfslide.net_em12c-capacity-planning-with-oem-metrics_2.pdf
pdfslide.net_em12c-capacity-planning-with-oem-metrics_2.pdfpdfslide.net_em12c-capacity-planning-with-oem-metrics_2.pdf
pdfslide.net_em12c-capacity-planning-with-oem-metrics_2.pdf
 
exadata-database-machine-kpis-3224944.pdf
exadata-database-machine-kpis-3224944.pdfexadata-database-machine-kpis-3224944.pdf
exadata-database-machine-kpis-3224944.pdf
 
TA110_System_Capacity_Plan.doc
TA110_System_Capacity_Plan.docTA110_System_Capacity_Plan.doc
TA110_System_Capacity_Plan.doc
 
kscope2013vst-130627142814-phpapp02.pdf
kscope2013vst-130627142814-phpapp02.pdfkscope2013vst-130627142814-phpapp02.pdf
kscope2013vst-130627142814-phpapp02.pdf
 
TGorman Collab16 UnixTools 20160411.pdf
TGorman Collab16 UnixTools 20160411.pdfTGorman Collab16 UnixTools 20160411.pdf
TGorman Collab16 UnixTools 20160411.pdf
 
BrownbagIntrotosqltuning.ppt
BrownbagIntrotosqltuning.pptBrownbagIntrotosqltuning.ppt
BrownbagIntrotosqltuning.ppt
 
sqltuningcardinality1(1).ppt
sqltuningcardinality1(1).pptsqltuningcardinality1(1).ppt
sqltuningcardinality1(1).ppt
 
ash-1-130820122404-phpapp01.pdf
ash-1-130820122404-phpapp01.pdfash-1-130820122404-phpapp01.pdf
ash-1-130820122404-phpapp01.pdf
 

Dernier

Subway Stand OFF storyboard by Raquel Acosta
Subway Stand OFF storyboard by Raquel AcostaSubway Stand OFF storyboard by Raquel Acosta
Subway Stand OFF storyboard by Raquel Acostaracosta58
 
The Hooper Talk (drama/comedy board sample)
The Hooper Talk (drama/comedy board sample)The Hooper Talk (drama/comedy board sample)
The Hooper Talk (drama/comedy board sample)DavonBrooks
 
Element of art, Transcreation and usions and overlapping and interrelated ele...
Element of art, Transcreation and usions and overlapping and interrelated ele...Element of art, Transcreation and usions and overlapping and interrelated ele...
Element of art, Transcreation and usions and overlapping and interrelated ele...jheramypagoyoiman801
 
UNIT 5-6 anh văn chuyên nganhhhhhhh.docx
UNIT 5-6 anh văn chuyên nganhhhhhhh.docxUNIT 5-6 anh văn chuyên nganhhhhhhh.docx
UNIT 5-6 anh văn chuyên nganhhhhhhh.docxssuser519b4b
 
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhool
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al MankhoolIndian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhool
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhoolqueenbanni425
 
Escort Service in Al Rigga +971509530047 UAE
Escort Service in Al Rigga +971509530047 UAEEscort Service in Al Rigga +971509530047 UAE
Escort Service in Al Rigga +971509530047 UAEvecevep119
 
Hiway Motel, Motel/Residence, Albuquerque NM
Hiway Motel, Motel/Residence, Albuquerque NMHiway Motel, Motel/Residence, Albuquerque NM
Hiway Motel, Motel/Residence, Albuquerque NMroute66connected
 
STAR Scholars Program Brand Guide Presentation
STAR Scholars Program Brand Guide PresentationSTAR Scholars Program Brand Guide Presentation
STAR Scholars Program Brand Guide Presentationmakaiodm
 
Escort Service in Al Qusais +971509530047 UAE
Escort Service in Al Qusais +971509530047 UAEEscort Service in Al Qusais +971509530047 UAE
Escort Service in Al Qusais +971509530047 UAEvecevep119
 
Value Aspiration And Culture Theory of Architecture
Value Aspiration And Culture Theory of ArchitectureValue Aspiration And Culture Theory of Architecture
Value Aspiration And Culture Theory of ArchitectureDarrenMasbate
 
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call Girls
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call GirlsAl Barsha Housewife Call Girls +971509530047 Al Barsha Call Girls
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call Girlshayawit234
 
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NM
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NMLindy's Coffee Shop, Restaurants-cafes, Albuquerque, NM
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NMroute66connected
 
Escort Service in Al Nahda +971509530047 UAE
Escort Service in Al Nahda +971509530047 UAEEscort Service in Al Nahda +971509530047 UAE
Escort Service in Al Nahda +971509530047 UAEvecevep119
 
New_Cross_Over (Comedy storyboard sample)
New_Cross_Over (Comedy storyboard sample)New_Cross_Over (Comedy storyboard sample)
New_Cross_Over (Comedy storyboard sample)DavonBrooks
 
Escort Service in Ajman +971509530047 UAE
Escort Service in Ajman +971509530047 UAEEscort Service in Ajman +971509530047 UAE
Escort Service in Ajman +971509530047 UAEvecevep119
 
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girls
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call GirlsAbu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girls
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girlshayawit234
 
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girl
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call GirlJvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girl
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girllijeho2176
 
Olympia Cafe, Restaurants-cafes, Albuquerque, NM
Olympia Cafe, Restaurants-cafes, Albuquerque, NMOlympia Cafe, Restaurants-cafes, Albuquerque, NM
Olympia Cafe, Restaurants-cafes, Albuquerque, NMroute66connected
 
Teepee Curios, Curio shop, Tucumcari, NM
Teepee Curios, Curio shop, Tucumcari, NMTeepee Curios, Curio shop, Tucumcari, NM
Teepee Curios, Curio shop, Tucumcari, NMroute66connected
 

Dernier (20)

Subway Stand OFF storyboard by Raquel Acosta
Subway Stand OFF storyboard by Raquel AcostaSubway Stand OFF storyboard by Raquel Acosta
Subway Stand OFF storyboard by Raquel Acosta
 
School :)
School                                 :)School                                 :)
School :)
 
The Hooper Talk (drama/comedy board sample)
The Hooper Talk (drama/comedy board sample)The Hooper Talk (drama/comedy board sample)
The Hooper Talk (drama/comedy board sample)
 
Element of art, Transcreation and usions and overlapping and interrelated ele...
Element of art, Transcreation and usions and overlapping and interrelated ele...Element of art, Transcreation and usions and overlapping and interrelated ele...
Element of art, Transcreation and usions and overlapping and interrelated ele...
 
UNIT 5-6 anh văn chuyên nganhhhhhhh.docx
UNIT 5-6 anh văn chuyên nganhhhhhhh.docxUNIT 5-6 anh văn chuyên nganhhhhhhh.docx
UNIT 5-6 anh văn chuyên nganhhhhhhh.docx
 
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhool
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al MankhoolIndian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhool
Indian Escorts In Al Mankhool 0509430017 Escort Agency in Al Mankhool
 
Escort Service in Al Rigga +971509530047 UAE
Escort Service in Al Rigga +971509530047 UAEEscort Service in Al Rigga +971509530047 UAE
Escort Service in Al Rigga +971509530047 UAE
 
Hiway Motel, Motel/Residence, Albuquerque NM
Hiway Motel, Motel/Residence, Albuquerque NMHiway Motel, Motel/Residence, Albuquerque NM
Hiway Motel, Motel/Residence, Albuquerque NM
 
STAR Scholars Program Brand Guide Presentation
STAR Scholars Program Brand Guide PresentationSTAR Scholars Program Brand Guide Presentation
STAR Scholars Program Brand Guide Presentation
 
Escort Service in Al Qusais +971509530047 UAE
Escort Service in Al Qusais +971509530047 UAEEscort Service in Al Qusais +971509530047 UAE
Escort Service in Al Qusais +971509530047 UAE
 
Value Aspiration And Culture Theory of Architecture
Value Aspiration And Culture Theory of ArchitectureValue Aspiration And Culture Theory of Architecture
Value Aspiration And Culture Theory of Architecture
 
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call Girls
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call GirlsAl Barsha Housewife Call Girls +971509530047 Al Barsha Call Girls
Al Barsha Housewife Call Girls +971509530047 Al Barsha Call Girls
 
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NM
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NMLindy's Coffee Shop, Restaurants-cafes, Albuquerque, NM
Lindy's Coffee Shop, Restaurants-cafes, Albuquerque, NM
 
Escort Service in Al Nahda +971509530047 UAE
Escort Service in Al Nahda +971509530047 UAEEscort Service in Al Nahda +971509530047 UAE
Escort Service in Al Nahda +971509530047 UAE
 
New_Cross_Over (Comedy storyboard sample)
New_Cross_Over (Comedy storyboard sample)New_Cross_Over (Comedy storyboard sample)
New_Cross_Over (Comedy storyboard sample)
 
Escort Service in Ajman +971509530047 UAE
Escort Service in Ajman +971509530047 UAEEscort Service in Ajman +971509530047 UAE
Escort Service in Ajman +971509530047 UAE
 
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girls
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call GirlsAbu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girls
Abu Dhabi Housewife Call Girls +971509530047 Abu Dhabi Call Girls
 
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girl
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call GirlJvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girl
Jvc Call Girl +971528604116 Indian Call Girl in Jvc By Dubai Call Girl
 
Olympia Cafe, Restaurants-cafes, Albuquerque, NM
Olympia Cafe, Restaurants-cafes, Albuquerque, NMOlympia Cafe, Restaurants-cafes, Albuquerque, NM
Olympia Cafe, Restaurants-cafes, Albuquerque, NM
 
Teepee Curios, Curio shop, Tucumcari, NM
Teepee Curios, Curio shop, Tucumcari, NMTeepee Curios, Curio shop, Tucumcari, NM
Teepee Curios, Curio shop, Tucumcari, NM
 

Oracle Wait Events That Everyone Should Know.ppt

  • 1. Oracle Wait Events That Everyone Should Know Kerry Osborne Senior Oracle Guy
  • 2.
  • 3. What Are Wait Events?  Basically a section of code  Uses gettimeofday  All time is stuck in a bucket with a name  Multiple things can be stuck in the same bucket  Oracle is very well instrumented (but It’s not 100%)  There are 41 classes of wait events in 10.2.0.3  There are 878 wait events in 10.2.0.3  209 enqueue events  29 latch events  41 I/O events
  • 4. What do you do with Wait Events?  See where Oracle is spending it’s time  Create a Resource Profile  Aggregate the time by Wait Event  Order the profile by time  Include # of occurrences and avg. time/event  Shows where to focus  Shows how much improvement can be made
  • 5. Profiles  Basic Tool – maybe the most important  Shows up in Statspack / AWR reports  Shows up in tkprof output (10g)  Can be pulled out of trace files  Can be pulled from V$ tables (v$system_event)  Can be pulled from ash tables Oracle is very well instrumented
  • 6. Profiles Top 5 Timed Events ~~~~~~~~~~~~~~~~~~ % Total Event Waits Time (s) Ela Time -------------------------------------------- ------------ ----------- -------- CPU time 37,297 52.27 SQL*Net message from dblink 7,065,802 11,312 15.85 async disk IO 943,852 5,265 7.38 db file sequential read 13,042,432 3,493 4.90 direct path write 108,862 3,410 4.78 -------------------------------------------------------------
  • 7. Why should developers know this stuff? 1. Because you need to know where your code is spending it’s time 2. Scalability is Important
  • 8.  1. These are the events you should know by heart  2. The other 260 or so you can figure out  3. You need to know what causes them  4. You need to know what values are reasonable  5. You need to know what you can do to fix them Top 10 List of Wait Events select name, parameter1, parameter2, parameter3 from v$event_name where name like nvl('&event_name',name) order by name
  • 9.  Not a wait event – but often included in a profile  Time spent but not accounted for in other buckets  Primarily time spent doing lio (we hope)  We want it to be on top  It may be an indicator of inefficient plans CPU
  • 10.  Single block read  Usually index block or data block via rowid  Can be done by fts to get chained rows  P1=file, P2=block, P3=# of blocks (always 1)  Always a user’s server process reading into buffer cache  Reasonable e values: <10ms DB File Sequential Read WAIT #14: nam='db file sequential read' ela= 36745 file#=1 block#=67745 blocks=1 obj#=84920
  • 11.  Do Less Work  tune SQL  reduce lio’s 7 / obj rule of thumb  Explain plan lies (see notes on this slide)  bigger buffer cache  Do the Work Faster  Speed up I/O  Call Jack DB File Sequential Read – “fixes”
  • 12.  Explain plan appears to executes a separate code path  Even if it didn’t there is no guarantee TEST matches PROD  Best bet is to execute and look at V$SQL_PLAN  This is easy in 10g with dbms_xplan select * from table(dbms_xplan.display_cursor('&sql_id','&child_no',''))  See an example in the notes section for this slide Digression – Explain Plan Lies
  • 13.  Multi block read  Usually full table scan or index fast full scan  P1=file, P2=block, P3=# of blocks (always < MBRC)  Always a user’s server process reading into buffer cache  Reasonable ela values: <10ms  Relevant parameters:  DBFMBRC, System Stats: MBRC,  _db_file_exec_read_count, _db_file_optimizer_read_count DB File Scattered Read WAIT #14: nam='db file scattered read' ela= 19389 file#=139 block#=44 blocks=5 obj#=83580
  • 14.  Do Less Work (fewer multi-block reads)  MBRC  Better Indexes  System Stats  bigger buffer cache  Tune SQL  Do the Work Faster  Speed up I/O DB File Scattered Read – “fixes”
  • 15.  Usually sorting to Temp  Can also be parallel query  Could also be insert append, etc…  Reasonable ela values: <20ms  Relevant parameters:  DBFMBRC  _db_file_direct_io_count (1M) Direct Path Read/Write WAIT #10: nam='direct path write' ela= 4475 p1=401 p2=1518353 p3=57 WAIT #10: nam='direct path read' ela= 16770 p1=401 p2=1482031 p3=63
  • 16.  Adjust PGA_AGGREGATE_TARGET  Turn off PX query  Call Randy Direct Path Read/Write – “fixes”
  • 17.  User process wait for LGWR to flush dirty buffers on commit  Synchronous write event  P1=log buffer block  Reasonable ela values: <4ms Log File Sync WAIT #16: nam='log file sync' ela= 1286 buffer#=11910 p2=0 p3=0 obj#=84920
  • 18.  Do Less Work (fewer commits)  Autocommit?  Row at a time processing with commits?  Do the Work Faster  Speed Up I/O  No RAID 5  No Contention (other db, arch, ASM)  SS Disk  Improve LGWR Scheduling  Renice Log File Sync – “fixes”
  • 19.  Wait for LGWR to write to current log files  Not necessarily synchronous write event  System I/O event  Reasonable ela values: <4ms  Occurs as follows:  Every 3 seconds  Log_buffer is 1/3 full or redo = 1M ( default _log_io_size)  Commit or rollback by user session  RAC needs to transfer a dirty block Log File Parallel Write
  • 20.  Often fixed by fixing Log File Sync (i.e. speed up i/o, etc)  Force LGWR to write more often Log File Parallel Write – “fixes”
  • 21.  Happens when database can’t switch to next log file  Freezes the whole database  Several Flavors  Checkpoint incomplete  Archiving needed  Reasonable ela values: 0ms Log file switch … WAIT #9: nam='log file switch (checkpoint incomplete)' ela= 986212 p1=0 p2=0 p3=0
  • 22.  Do less work  Make total online log space bigger  Force more frequent incremental checkpoints  Speed up i/o Log file switch … – “fixes”
  • 23.  Contention event for the same block  Read by other session (new event in 10g)  Held by other session in incompatible mode  New event in 10g – read by other session  Multiple sections of code throw time in this bucket  P1=file, P2=block, P3=reason code  See metalink note (34405.1) for details Buffer Busy Waits WAIT #7: nam='read by other session' ela= 3251 file#=4 block#=188557 class#=1 obj#=53707 tim=1183096831514887
  • 24.  It’s complicated – but basically - eliminate the contention  For Read by Other Session  Do Less Work (tune SQL to do less lio)  Eliminate i/o – bigger buffer cache, etc…  Speed up i/o  For Others  Find type of hot objects and statements suffering  Address issue - ASSM, Freelist groups (RAC), etc.. Buffer Busy Waits - “fixes”
  • 25.  No place to put a new block Free Buffer Waits
  • 26.  Do Less Work  Reduce the amount of lio  Make more buffer space available  Add memory to the buffer cache  Speed up DBWR to flush blocks more quickly  Async_io  More dbwr processes  Renice Free Buffer Waits - “fixes”
  • 27.  Database is idle – waiting on next request from client  Often marked as an idle event which it may be  Time distribution can be skewed  Common technique to ignore anything over 1 sec  Reasonable e values: <2ms  Protocol should be tcp (or beq for local processes) SQL*Net message from client WAIT #1: nam='SQL*Net message from client' ela= 336 driver id=1650815232 #bytes=1 p3=0 obj#=83660
  • 28.  Speed up app server  Reduce the number of calls  Use the right protocol (tcp to beq)  Call network guy  Get the users to type faster SQL*Net message from client – “fixes”
  • 29.  More than 1 packet required to send SQL statement SQL*Net more data from client
  • 30.  Don’t send 50K SQL statements  Use packages SQL*Net more data from client – “fixes”
  • 31.  Time to send a SQL*Net message to a client (sort of)  Actually time it took to write the return data from Oracle’s userland SDU buffer to OS kernel-land TCP socket buffer (Tanel Poder)  Often marked as an idle event which it may be  Reasonable e values: <1ms (micro seconds actually)  P1=bytes (usually 1 ???) SQL*Net message to Client WAIT #1: nam='SQL*Net message to client' ela= 2 driver id=1650815232 #bytes=1 p3=0 obj#=83660
  • 32.  Use correct protocol  Fix network issue  Call Andy SQL*Net message to Client – “fixes”
  • 33.  When more than one packet required  Rows spanning block  LOB’s  Array fetches  Often marked as an idle event which it may be  Reasonable e values: <1ms  P1=bytes (usually 1 ???) SQL*Net more data to client WAIT #1: nam='SQL*Net message to client' ela= 2 driver id=1650815232 #bytes=1 p3=0 obj#=83660
  • 34.  Fix chained/migrated rows  Don’t use LOBs  Nothing to fix SQL*Net more data to client – “fixes”
  • 35.  Locks for objects  Que up in order  9i lumped them all together  10g has 209 separate events Enqueues
  • 36.  P1 has encoded type and mode Enqueue in 9i SELECT chr(bitand(p1,-16777216)/16777215)|| chr(bitand(p1, 16711680)/65535) Type, mod(p1,16) lmode from v$session_wait where event=‘enqueue’;
  • 37.  Several Major Categories  TM – table modification  TX – Transaction locks  UL – user lock  CI – Cross Instance  CU – Cursor Bind  HW – High Water  RO – Reusable Object  ST – Space Transaction  TS – Temporary Space  Parameters depend on type Enqueue in 10g
  • 38.  User can’t modify block until other user commits Enqueue: TX row lock
  • 39.  Use ash tables to find statements waiting  Fix the code  Or do what one of our clients did: write some code to kill any processes blocking for more than 60 seconds – lot’s easier than changing the app Enqueue: TX row lock – “fixes”
  • 40.  Waiting on exclusive access to a sequence  Seems minor but can actually be a big issue on some systems Enqueue: SQ
  • 41.  Cache the sequence  Use NOORDER in RAC Enqueue: SQ – “fixes”
  • 42.  Locks internal memory structures  Not ordered  Very light weight  Kid Asking for Candy Model  Spins vs. Sleeps (_spin_count, _latch_class_X)  Wait events do not include time spent spinning  9i lumped them all together  10g has 28 separate events Latches
  • 43.  All dumped in the Latch Free event  P2 has latch# Latches in 9i WAIT #9: nam='latch free' ela= 185 p1=-4611686003650090584 p2=157 p3=0 SQL> Select name from v$latch where latch#=157; NAME ---------------------------------------- library cache
  • 44.  Primarily Deal with major SGA areas  Shared Pool  Latch: shared pool  Latch: library cache  Latch row cache  Buffer Cache  Latch: cache buffer chains  Latch: cache buffers lru chain  Log Buffer  Latch: redo …  Parameters depend on type (see v$event_name) Latches in 10g
  • 45.  Each block hashes to a chain  Several chains per latch  Several latches per instance (default based on cache size) Latch: cache buffers chains
  • 46.  Find hot blocks  Fix bad SQL  Maybe Increase number of latches  Maybe decrease rows per block Latch: cache buffers chains –”fixes”
  • 47.  Shared Pool Contention  Due primarily to parsing  See also latch: library cache … Latch: shared pool WAIT #28: nam='latch: shared pool' ela= 751 address=1611562656 number=213 tries=1 j#=46024
  • 48.  Use Bind Variables  CURSOR_SHARING = FORCE  Increase SHARED_POOL – this is where Darcy came in and sat in my office so I didn’t get much more done – You can thank Darcy later! Latch: shared pool – “fixes”
  • 49.  Some tools show this (Hotsos Profiler)  Indicates lost time due to rounding (small)  Or lost time due to CPU contention (large) Unaccounted for
  • 50. Tracing  alter session set events '10046 trace name context forever, level 8';  Spits out wait events, plans (real), stats  Level 12 includes bind variables – makes file very big  Puts file in user_dump_dest (ls –altr)  Can be executed from logon trigger, in-line, etc…  Scoping is important (i.e. when it’s turned on and off)  Note: bug in early versions of 9.2, fixed in 9.2.0.6
  • 51. Trace File Contents  WAIT  FETCH  EXEC  PARSE  STAT  XCTEND  PARSING IN CURSOR
  • 52. Raw Trace File  /opt/oracle/admin/LAB102/udump/lab102_ora_4207.trc  Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Production  ORACLE_HOME = /opt/oracle/product/db/10.2.0/db1  System name: Linux  Node name: homer  Release: 2.6.9-34.ELhugemem  Version: #1 SMP Fri Feb 24 17:04:34 EST 2006  Machine: i686  Instance name: LAB102  Redo thread mounted by this instance: 1  Oracle process number: 20  Unix process pid: 4207, image: oracle@homer (TNS V1-V3)  *** ACTION NAME:() 2007-08-16 13:48:14.571  *** MODULE NAME:(SQL*Plus) 2007-08-16 13:48:14.571  *** SERVICE NAME:(SYS$USERS) 2007-08-16 13:48:14.571  *** SESSION ID:(143.189) 2007-08-16 13:48:14.571  =====================  PARSING IN CURSOR #7 len=68 dep=0 uid=61 oct=42 lid=61 tim=1159462982979740 hv=740818757 ad='30663e48'  alter session set events '10046 trace name context forever, level 8'  END OF STMT  EXEC #7:c=0,e=269,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=1,tim=1159462982979717  WAIT #7: nam='SQL*Net message to client' ela= 7 driver id=1650815232 #bytes=1 p3=0 obj#=-1 tim=1159462982980778  WAIT #7: nam='SQL*Net message from client' ela= 119 driver id=1650815232 #bytes=1 p3=0 obj#=-1 tim=1159462982981008  =====================  PARSING IN CURSOR #8 len=44 dep=0 uid=61 oct=3 lid=61 tim=1159463023994427 hv=761757617 ad='54738434'  select avg(col1) from skew  where rownum < 10  END OF STMT  PARSE #8:c=4000,e=3904,p=0,cr=0,cu=0,mis=1,r=0,dep=0,og=1,tim=1159463023994411  EXEC #8:c=0,e=185,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=1,tim=1159463023994844  WAIT #8: nam='SQL*Net message to client' ela= 9 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023994980  WAIT #8: nam='db file scattered read' ela= 218 file#=4 block#=21900 blocks=5 obj#=53707 tim=1159463023995544  FETCH #8:c=0,e=665,p=5,cr=4,cu=0,mis=0,r=1,dep=0,og=1,tim=1159463023995732  WAIT #8: nam='SQL*Net message from client' ela= 157 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996048  FETCH #8:c=0,e=7,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=0,tim=1159463023996183  WAIT #8: nam='SQL*Net message to client' ela= 7 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996312  WAIT #8: nam='SQL*Net message from client' ela= 298 driver id=1650815232 #bytes=1 p3=0 obj#=53707 tim=1159463023996669  XCTEND rlbk=0, rd_only=1  STAT #8 id=1 cnt=1 pid=0 pos=1 obj=0 op='SORT AGGREGATE (cr=4 pr=5 pw=0 time=736 us)'  STAT #8 id=2 cnt=9 pid=1 pos=1 obj=0 op='COUNT STOPKEY (cr=4 pr=5 pw=0 time=846 us)'  STAT #8 id=3 cnt=9 pid=2 pos=1 obj=53707 op='TABLE ACCESS FULL SKEW (cr=4 pr=5 pw=0 time=639 us)'  WAIT #0: nam='log file sync' ela= 680 buffer#=4862 p2=0 p3=0 obj#=53707 tim=1159463039852003