Legacy HDS Forums

pfm files statistics

Discussion created by Legacy HDS Forums on Jul 27, 2009
Latest reply on Aug 11, 2009 by Legacy HDS Forums

Originally posted by: przemolb



I have been logging all performance-related information from our arrays using pfm* files.
Each minute is in separated file. I took one of those files:
[code]
No.856
2009/07/24 14:15:05 - 2009/07/24 14:16:04
[/code]
and counted IOPS which goes to ports:
[code]
---- Port Information ----
CTL Port IO Rate(IOPS) Read Rate(IOPS) Write Rate(IOPS) Read Hit(%) Write Hit(%) Trans. Rate(MB/S) Read Trans. Rate(MB/S) Write Trans. Rate(MB/S)
0 A 54 23 30 94 100 1 0 0
0 B 2076 1202 874 51 100 62 38 23
1 A 52 23 28 97 100 1 0 0
1 B 1486 1111 374 52 100 40 35 5
[/code]
sum: 3412

Then I counted all IOPS from the same file which are reported to get to HDDs:
sum: 2359

It seems that either IOPS are merged or some of them get to disks after the end of reported time (14:16:04).
What is true ?

Outcomes