- Community Home
- >
- Servers and Operating Systems
- >
- Operating Systems
- >
- Operating System - HP-UX
- >
- Re: Oracle DB and Kernel Parameters
Operating System - HP-UX
1754365
Members
4415
Online
108813
Solutions
Forums
Categories
Company
Local Language
юдл
back
Forums
Discussions
Forums
- Data Protection and Retention
- Entry Storage Systems
- Legacy
- Midrange and Enterprise Storage
- Storage Networking
- HPE Nimble Storage
Discussions
Discussions
Discussions
Forums
Forums
Discussions
юдл
back
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
- BladeSystem Infrastructure and Application Solutions
- Appliance Servers
- Alpha Servers
- BackOffice Products
- Internet Products
- HPE 9000 and HPE e3000 Servers
- Networking
- Netservers
- Secure OS Software for Linux
- Server Management (Insight Manager 7)
- Windows Server 2003
- Operating System - Tru64 Unix
- ProLiant Deployment and Provisioning
- Linux-Based Community / Regional
- Microsoft System Center Integration
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Blogs
Information
Community
Resources
Community Language
Language
Forums
Blogs
Go to solution
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО02-25-2003 12:58 AM
тАО02-25-2003 12:58 AM
Re: Oracle DB and Kernel Parameters
Hi Ketil,
I think the proble you have results from having only one disk in your HP box.
Making a copy of a large file does not realy show how fast your disk is because you 'll see only the speed for a sequential read/write operation. But in fact, as Oracle does a lot of seeks before read/write a striped diskset will be much faster than a single disk ! I had the same problems here having a warehouse system running on several different systems (at different locations). And even 2 HP boxes which differes only in the harddisk configuration ran at different performance. Same issue. You can do now:
1. Add more disks (only for the oracle db, leave the Unix system on the first one) and stripe them if money doesn't matter
2. Have a look on the memory statistics. If there's a plenty of memory free when running the system on full load you might increase the kernels i/o buffer size (normally set to min 5% and max 50% (dynamical) with dbc_min_pct/dbc_mac_ptc) to minimize direct reads/writes.
3. Also have a look on the Oracle DB_cache (db_block_buffers). If theres free memory in your Shared memory area free, try if increasing that might help.
4. Are you using asyncron i/o ?
5. Check the log-buffer statistics. Sometimes increasing the log-buffer helps a little bit. Also check how often logfiles are writen.
6. Does your application make a lot of sorts? If then check if there are any disks sorts. Try to get rid of them by increasing the sort_area_size.
Maybe that might help you.
Regards,
Michael
I think the proble you have results from having only one disk in your HP box.
Making a copy of a large file does not realy show how fast your disk is because you 'll see only the speed for a sequential read/write operation. But in fact, as Oracle does a lot of seeks before read/write a striped diskset will be much faster than a single disk ! I had the same problems here having a warehouse system running on several different systems (at different locations). And even 2 HP boxes which differes only in the harddisk configuration ran at different performance. Same issue. You can do now:
1. Add more disks (only for the oracle db, leave the Unix system on the first one) and stripe them if money doesn't matter
2. Have a look on the memory statistics. If there's a plenty of memory free when running the system on full load you might increase the kernels i/o buffer size (normally set to min 5% and max 50% (dynamical) with dbc_min_pct/dbc_mac_ptc) to minimize direct reads/writes.
3. Also have a look on the Oracle DB_cache (db_block_buffers). If theres free memory in your Shared memory area free, try if increasing that might help.
4. Are you using asyncron i/o ?
5. Check the log-buffer statistics. Sometimes increasing the log-buffer helps a little bit. Also check how often logfiles are writen.
6. Does your application make a lot of sorts? If then check if there are any disks sorts. Try to get rid of them by increasing the sort_area_size.
Maybe that might help you.
Regards,
Michael
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО02-25-2003 01:23 AM
тАО02-25-2003 01:23 AM
Re: Oracle DB and Kernel Parameters
Hi
Do you have access to metalink
Check the patches list
and the following note :
Doc ID: Note:200075.1
Subject: Oracle 9iAS Release 2 (9.0.2.0.1) Installation Cookbook
Type: BULLETIN
Status: PUBLISHED
Content Type: TEXT/PLAIN
Creation Date: 18-JUN-2002
Last Revision Date: 20-FEB-2003
Do you use the same parameter in the two database :
yau can check the parameter in the base with the following order :
select * from v$paramter;
Do you use the same repartition on physical disk ? (data, system, index)
Do you have made a compute statistic ?
Do you have access to metalink
Check the patches list
and the following note :
Doc ID: Note:200075.1
Subject: Oracle 9iAS Release 2 (9.0.2.0.1) Installation Cookbook
Type: BULLETIN
Status: PUBLISHED
Content Type: TEXT/PLAIN
Creation Date: 18-JUN-2002
Last Revision Date: 20-FEB-2003
Do you use the same parameter in the two database :
yau can check the parameter in the base with the following order :
select * from v$paramter;
Do you use the same repartition on physical disk ? (data, system, index)
Do you have made a compute statistic ?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО02-25-2003 01:27 AM
тАО02-25-2003 01:27 AM
Re: Oracle DB and Kernel Parameters
Thanks for all advices.
We are currently installing more disks into your server, so we are hoping for a good increase in performance.
Compared to our Windows system, especially large inserts / updates are relatively slower. From "sar -d" I see that %busy is continously at 100%, so it seems that disk-speed is a limiting factor compared to the windows system, since init.ora is approx. the same in the two systems.
Ketill
We are currently installing more disks into your server, so we are hoping for a good increase in performance.
Compared to our Windows system, especially large inserts / updates are relatively slower. From "sar -d" I see that %busy is continously at 100%, so it seems that disk-speed is a limiting factor compared to the windows system, since init.ora is approx. the same in the two systems.
Ketill
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО02-25-2003 02:37 AM
тАО02-25-2003 02:37 AM
Re: Oracle DB and Kernel Parameters
hello,
At the same it would recommended that you move to locally managed tablespaces.
Amongst advantages:
o no fragmentation
o faster (no recursive sql to UET$ and FET$ to update the data dictionary)
o faster (try creating a dictionary managed table with 1000 extents, then do it with a LMT)
o faster (try dropping that table with 1000 extents now)
o faster (try a temporary tablespace in a DMT vs LMT -- benchmark it, its amazing)
About Rollback segments:
Place individual rollback segments into their own tablespaces.
You like to start with 15 or so rollback segments -- each with minextents=25 and maxextents=100. That would create 15 25meg rollback segments that can grow to 100meg each.
create tablespace RBS_01
datafile '/xxx/xxx/xxx/rbs_01.dbf'
size 25m reuse
autoextend on
maxsize 100m
extent management local uniform size 1m;
Then:
Create rollback segment rbs_01 tablespace rbs_01;
alter rollback segment rbs_01 online;
where:
o rollback segments are equi-sized extents.
o rollback segments are allocated large to avoid 1555's.
o rollback segments should be plentiful for the amount of transactions you do.
Hope this helps!
Best Regards
Yogeeraj
At the same it would recommended that you move to locally managed tablespaces.
Amongst advantages:
o no fragmentation
o faster (no recursive sql to UET$ and FET$ to update the data dictionary)
o faster (try creating a dictionary managed table with 1000 extents, then do it with a LMT)
o faster (try dropping that table with 1000 extents now)
o faster (try a temporary tablespace in a DMT vs LMT -- benchmark it, its amazing)
About Rollback segments:
Place individual rollback segments into their own tablespaces.
You like to start with 15 or so rollback segments -- each with minextents=25 and maxextents=100. That would create 15 25meg rollback segments that can grow to 100meg each.
create tablespace RBS_01
datafile '/xxx/xxx/xxx/rbs_01.dbf'
size 25m reuse
autoextend on
maxsize 100m
extent management local uniform size 1m;
Then:
Create rollback segment rbs_01 tablespace rbs_01;
alter rollback segment rbs_01 online;
where:
o rollback segments are equi-sized extents.
o rollback segments are allocated large to avoid 1555's.
o rollback segments should be plentiful for the amount of transactions you do.
Hope this helps!
Best Regards
Yogeeraj
No person was ever honoured for what he received. Honour has been the reward for what he gave (clavin coolidge)
- « Previous
-
- 1
- 2
- Next »
The opinions expressed above are the personal opinions of the authors, not of Hewlett Packard Enterprise. By using this site, you accept the Terms of Use and Rules of Participation.
News and Events
Support
© Copyright 2024 Hewlett Packard Enterprise Development LP