Unsolved
This post is more than 5 years old
5 Practitioner
•
274.2K Posts
0
1044
Host Performance issues
Hi Guys!
Can any one help me with some performance questions? Symetrix 8830
I am in a situation where we have staff in "DataBase" who report performance issues.
They then go to "OS", Solaris in this instance, who immediatly throw it over the wall at us "SAN".
I would like to put in place some sort of triage where they at least have to run sar or iostat and give us reasons why they think it is the disk etc.
I am thinking they should check for diskbusy%, disk queue length and io response.
Do any of you have any suggestions as to any more metrics I can ask them to check or any I can get the dba's to check?
Thanks in advance.
Dave
Can any one help me with some performance questions? Symetrix 8830
I am in a situation where we have staff in "DataBase" who report performance issues.
They then go to "OS", Solaris in this instance, who immediatly throw it over the wall at us "SAN".
I would like to put in place some sort of triage where they at least have to run sar or iostat and give us reasons why they think it is the disk etc.
I am thinking they should check for diskbusy%, disk queue length and io response.
Do any of you have any suggestions as to any more metrics I can ask them to check or any I can get the dba's to check?
Thanks in advance.
Dave
sscss0
15 Posts
1
November 27th, 2006 10:00
I asked our Oracle DBA guy and he said there is a tool in Oracle called Stats Pack. He said to run that and look at the top 5 wait events for that database.
This will show exactly what Oracle is waiting on and causing delays. Also check out any possible network problems. I rarely find that the disk backend is a problem.
sysmgr1
128 Posts
0
November 28th, 2006 12:00
The best bet is to keep them busy with something else. There is a module called OEM which is a suite of performance and monitoring tools that can give them a number of good performance stats. However, the %disk busy, disk queue length, io response time, and the iowait (vmstat -w) are sufficient to keep them bugging you for years.
Orrinf
2 Posts
0
December 13th, 2006 18:00
GlenH
141 Posts
0
December 17th, 2006 19:00
There is a very good, detailed explanation of the Solaris disk performance statistics here: http://storage.itworld.com/4650/UIR990801perf/pfindex.html which is definitely worth a read.
As a general rule, the database log files are the most critical when it comes to response times - the data files generally don't care so much as you usually want high throughput for data files which is best achieved when you issue multiple I/O's to each device causing a higher respose time...
If it's Oracle, then definately get your DBA to look at the Statspack information first so they can give you the cause of the performacne issue - that will give you the best indication as to why the database is slowing down and allow you to narrow the search down considerably.
I would also recommend that both you and your DBA take a look at Steve Adam's ixora site here: http://www.ixora.com.au/tips/ which is an excellent resource that has Oracle performance best-practices, layouts & tuning information.
Good Luck,
Glen.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
January 7th, 2007 08:00
Regards,
Dave