I have uploaded HTnM output(IOPS and IO response time) for 2 Ldevs from one of our linux server. These two ldevs are giving performance issues.
Please let me know if the output is ok or not.
Are the performance issues happening all the time?
Looks like decent response times to me. You do have the few spikes, but if the customer is complaining about performance all the time, then I wouldn't worry about the spikes. If, however, the customer is complaining only during the duration of the spikes, then you have a problem.
I personally consider anything under 15ms non-problematic.
If they are complaining about performance all the time, then it's most likely a bottleneck elsewhere.
What could be a reasonable response time..
here in my case. it is showing 100msec. What could be a alraming level.
Also IOPS is approx 1500..
Is it ok..
TIll what level we can say stable level.
That piece of information is not enough to provide any help to you. :-(
What is the exact problem statement? Which storage array you have?
how is subsystem MP utilization and CWP doing?
How many hosts you had on that particular storage port sharing the bandwidth with this linux server? What disk, raid, pool config you have in that array?
There could be lot of factors and there is a possibility that issue could be elsewhere.
We have one linux filesystem made of these two ldevs, where the DB team is facing slow performance.
Raid 6 sas disk DP pool
How big are the disk?
What type of DB?
Ldevs are of 200GB each.
I suggest you keep your oracle archives, and rman backups together on one set of disk, your redo logs on another disk, and your database data files on another.
So maybe for example:
30 GB disk for Redo logs
100 GB for RMAN/Archives
200 for data files
The above is just an example of splitting out your disk by function. Work with the database administrator for actual sizes.
Putting all that on the same set of disks will potentially cause performance issues.
Retrieving data ...