Login or Register to Ask a Question and Join Our Community


ORACLE Database running slow on AIX ( nmon / topas )


 
Thread Tools Search this Thread
Operating Systems AIX ORACLE Database running slow on AIX ( nmon / topas )
# 36  
Old 11-11-2010
First of all, you should lower your SGA i would set it to 20Gb for a beginning and put it in memory with lock_sga = true
You should lower the db_cache_size as well (must be contained in sga), 32G is too big regarding to the current activity which show around 620M to be currently sufficient ....
so setting a db_cache_size to 2Gb would already be quite generous

it was lock_sga, not log_sga (that was a typo error that i have corrected)

You should consider setting lock_sga to true instead of the current value false, but make sur you fit the oracle recommendation (see the previous pdf and metalink reference previously posted)

I did a typo error : in the view name of the PGA : it is v$pgastat instead of v$pgastats (no trailing "s")
Code:
select * from v$pgastat;

By the way , you might consider setting your lock_sga to true instead of false since it seems you have enough physical memory to hold your SGA into it (this will force the SGA in RAM and not in swap). But do it carefully (if all the RAM is used by the SGA, there will be less RAM for PGA ...) and make sure you follow the oracle recommendations in the pdf link mentionned in my previous post.

Maybe you should generate an html report on your server using the standard report, @$ORACLE_HOME/rdbms/admin/awrrpt.sql
and selecting 2 snaps containing a timeframe in which you encountered performance problem.(The timeframe should not contain downtime, otherwise the reported stat are not relevant) so you could then have a look at it (have a look at heavy sql statements for example).

Does your server run a single oracle instance or more ?

Check how much RAM is used purely by your operating system.

Do you have some other application or ERP running on that server (SAP ? OAP ? other ?) or is it a pure oracle server?

Last edited by ctsgnb; 11-11-2010 at 09:55 PM..
# 37  
Old 11-11-2010
The only reason your box is still alive is that the SGA is not pinned into memory if it is so huge. If you even try to do that the box will no doubt crash after some uptime as the kernel is pinning memory too and you cannot exceed the overall possible value of pinned memory that is about 85%. With sufficient tuning and sufficient memory, AIX is not going to page computational memory as long as the avm value doesnt exceed 97% of total physical memory. Btw I mentioned it before - your minperm is 10% - even by default it is on AIX 6.1 3% and with your memory consumption on the box you should do the same even for AIX 5.3.
# 38  
Old 11-11-2010
The SGA is OVERsized ...
as well as the db_cache_size which is told not to decrease below 32G whereas it seems 75328 x 8192 ~ 620M are currently enough ...

Last edited by ctsgnb; 11-11-2010 at 09:09 AM..
These 2 Users Gave Thanks to ctsgnb For This Post:
# 39  
Old 11-11-2010
Any chance of seeing the initSID.ora file ?
I agree that say 24 Gb total SGA (25% of memory) would be generous unless this database has a history of ORA errors with "normal" values.
An excessively large db_cache_size can itself cause performance issues on databases with a high rate of data change.
# 40  
Old 11-11-2010
A reason that would justify such a big SGA would be if we would have plenty of users and the server would run in shared server mode (the UGA is then located in SGA instead of the PGA).

But it doesn't seem to be our case.
Please give us the output of
Code:
show parameter share

... just to make it sure
# 41  
Old 11-12-2010
initPROD.ora
Code:
_b_tree_bitmap_plans = FALSE
_enable_shared_pool_durations = FALSE
_fast_full_scan_enabled = FALSE
_like_with_bind_as_equality = TRUE
_sort_elimination_cost_ratio = 5
_sqlexec_progression_cost = 2147483647
_system_trig_enabled = TRUE
_trace_files_public = TRUE
aq_tm_processes = 5
background_dump_dest = /u03/oracle/PROD/db/tech_st/10.2.0/admin/PROD_nmcretek01/bdump
compatible = 10.2.0
control_files = ('/u03/oracle/PROD/db/apps_st/data/cntrl01.dbf', '/u03/oracle/PROD/db/apps_st/data/cntrl02.dbf', '/u03/oracle/PROD/db/apps_st/data/cntrl03.dbf')
core_dump_dest = /u03/oracle/PROD/db/tech_st/10.2.0/admin/PROD_nmcretek01/cdump
cursor_sharing = EXACT
db_block_checking = FALSE
db_block_checksum = TRUE
db_block_size = 8192
db_cache_size = 34359738368
db_files = 512
db_name = PROD
dml_locks = 10000
job_queue_processes = 10
log_archive_dest_1 = 'LOCATION=/u06/oracle/ARCHIVES/PROD_ARCHIVES'
log_archive_format = 'PROD_%s_%t_%r.arc'
log_buffer = 62906368
log_checkpoint_interval = 100000
log_checkpoint_timeout = 1200
log_checkpoints_to_alert = TRUE
max_dump_file_size = 20480
nls_comp = binary
nls_date_format = DD-MON-RR
nls_language = american
nls_length_semantics = BYTE
nls_numeric_characters = '.,'
nls_sort = binary
nls_territory = america
olap_page_pool_size = 4194304
open_cursors = 4000
optimizer_secure_view_merging = FALSE
parallel_max_servers = 12
parallel_min_servers = 0
pga_aggregate_target = 6442450944
plsql_code_type = INTERPRETED
plsql_native_library_dir = /u03/oracle/PROD/db/tech_st/10.2.0/plsql/nativelib
plsql_native_library_subdir_count = 149
plsql_optimize_level = 2
processes = 1200
recyclebin = off
session_cached_cursors = 500
sessions = 2400
sga_target = 85899345920
shared_pool_reserved_size = 629145600
shared_pool_size = 6442450944
timed_statistics = TRUE
undo_management = AUTO
undo_tablespace = APPS_UNDOTS1
user_dump_dest = /u03/oracle/PROD/db/tech_st/10.2.0/admin/PROD_nmcretek01/udump
utl_file_dir = ('/usr/tmp','/u03/oracle/PROD/db/tech_st/10.2.0/appsutil/outbound/PROD_nmcretek01','/u03/oracle/PROD/apps/apps_st/comn/temp','/usr/tmp')
workarea_size_policy = AUTO

and

Code:
SQL> show parameter lock_sga;

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
lock_sga                             boolean     FALSE

I will take the recommendations and will make the changes and then post the results...
# 42  
Old 12-12-2010
The DBA has made some changes as requested by the expert posters in this thread. I will soon post the output as there is some notable difference in the performance.

However, one question to ZXMaus:
You requested to change
Your minfree is too high, set it to 3. Your minperm is 10% - even by default it is on AIX 6.1 3%

So, should I set both MINFREE and MINPERM or just one of them to 3%

vmo -p -o minperm%=3
vmo -p -o minfree=3

Thanks
Login or Register to Ask a Question

Previous Thread | Next Thread

8 More Discussions You Might Find Interesting

1. AIX

AIX server running very slow

Hello, All the commands on AIX are running very slow. Below is few stats but I didn't find any issue in cpu or memory reosurces vmstat System configuration: lcpu=4 mem=6144MB ent=1.00 kthr memory page faults cpu ----- -----------... (2 Replies)
Discussion started by: Vishal_dba
2 Replies

2. AIX

Need to check long running processes on the database server and the os is AIX

Hello, Please help me with a script with which I can check long running processes on the database server and the os is AIX. Best regards, Vishal (5 Replies)
Discussion started by: Vishal_dba
5 Replies

3. Shell Programming and Scripting

Running script automatically when threshold limit met in one of the field in oracle database

Hi Guys, Need you help in one point! I am working on one shell script which takes following steps : 1. Taking one query result from oracle database 2. Exporting that result to Xls file 3. Mailing that file to my own mail ID Now, I want to give a threshold limit to one of the column... (0 Replies)
Discussion started by: Agupte
0 Replies

4. Solaris

what is the command to see which database ie..oracle in running on solaris 10

what is the command to see which database ie..oracle in running on solaris 10 (2 Replies)
Discussion started by: tv.praveenkumar
2 Replies

5. AIX

Top command in AIX 4.2 (no topas, no nmon, no top)?

Is there a 'top' command equivalent in AIX 4.2 ? I already checked and I do not see the following ones anywhere: top nmon topas (1 Reply)
Discussion started by: Browser_ice
1 Replies

6. AIX

nmon vs topas

good morning what is the better solution to examen a P570 ? because i use topas and nmon, and the results are totally different !!! with nmon, i have 80% free cpu, and with nmon, i have 90% of used cpu !!!!!! i take a shot with an intervall of 10s during 10 mn. thank you (0 Replies)
Discussion started by: pascalbout
0 Replies

7. Filesystems, Disks and Memory

skgpspawn failed running oracle db 9.2.0.5.0 on aix 5.3

Hi, I am running an oracle db 9.2.0.5.0 on ibm p5 550 aix 5.3 with 10g ram, 10G swap space 3 database instances each SGA about 500Meg. I am getting the following error in my alert log file from time to time: skgpspawn failed:category = 27142, depinfo = 11, op = fork, loc = skgpspawn3 ... (0 Replies)
Discussion started by: hawkerpacific
0 Replies

8. AIX

Memory usage statistic? (topas, nmon)

hi, how can i diplay: - the ammount RAM used /free - ammount of ram used from a pid or prozess we have the problem, that malloc is returing a NULL pointer errno = 12 ( not enough space). but i think there is still ram free. nmon : shows all memory used ? Memory Use Physical Virtual... (7 Replies)
Discussion started by: Lazzar
7 Replies
Login or Register to Ask a Question

Featured Tech Videos