Interpret cluster-wide Disk Statistics

Evaluate Performance in Detail › Performance Evaluation Report › Interpret cluster-wide Disk Statistics

Interpret cluster-wide Disk Statistics

The disk statistics follow the cluster-wide CI, NI, and adapter statistics. To display only the disk statistics section of the Performance Evaluation Report, use the following qualifier:

/INCLUDE=DISK_STATISTICS

To disable the disk statistics display from the report, use the following qualifier:

/INCLUDE=NODISK_STATISTICS

The following report shows an example of the disk statistics section of the Performance Evaluation report:

Performance CLUSTER Page 8 Evaluation PA Vx.x Wednesday 14-JAN-2006 13:00 to 14:00 +-----------------------------------------------------------------------+ | The following table gives the summary of all disk activity as seen | | by the indicated node. An "*" for service node indicates that more | | than one was detected. | +-----------------------------------------------------------------------+ Disk Avg I/O Avg Avg IOsz Source Service % % IO % IO # of Volume per Sec Queue Kb/sec in pgs Node Node Busy Read Split Type Samples ------ ------- ----- ------ ------ ------ ------ ------ ---- ----- ---- ------- 1 2 3 4 5 6 7 8 9 10 11 12 ARNOLD ($2$DJA4) 0.00 0.00 0.0 0.0 NODE1 0 0 RA60 0.00 0.00 0.0 0.0 SUPPLY 0.00 0 0 30 0.00 0.00 0.0 0.0 DEMAND 0.00 0 0 30 0.00 0.00 0.0 0.0 NODE1 0.00 0 0 30 BARRY ($2$DUA15) 0.66 0.03 2.3 7.0 ERNIE 100 4 RA81 0.00 0.00 0.0 0.0 SUPPLY 0.00 0 0 30 0.66 0.03 2.3 7.0 DEMAND 1.97 100 4 30 0.00 0.00 0.0 0.0 NODE1 0.00 0 0 30 CLEM ($2$DUA5) 2.02 0.06 5.0 4.9 ERNIE 83 1 RA82 0.14 0.00 0.3 3.6 SUPPLY 0.35 98 0 30 1.46 0.04 3.8 5.2 DEMAND 3.89 82 2 30 0.42 0.01 1.0 4.6 NODE1 1.15 81 0 30 DUFIS ($2$DUA10) 3.86 0.14 16.5 8.6 BERT 86 15 RA81 0.00 0.00 0.0 0.0 SUPPLY 0.00 0 0 30 1.98 0.08 9.1 9.2 DEMAND 7.31 84 13 30 1.88 0.06 7.4 7.9 NODE1 5.66 89 17 30 ELLIE ($2$DUA7) 8.93 0.40 28.5 6.4 BERT 93 0 RA82 0.04 0.01 1.6 74.4 SUPPLY 0.26 15 0 30 8.36 0.35 22.0 5.3 DEMAND 18.73 95 0 30 0.53 0.04 5.0 19.0 NODE1 2.15 79 0 30

The following statements are keyed to the columns in the previous example:

Volume name of the disk to which one or more nodes in the cluster directs activity.
Average number of I/O operations per second to the disk volume. For each disk, the total I/O per second is reported, followed by a breakdown of this activity from contributing nodes, when applicable. In this example, there was an average of 2.02 I/O operations per second to disk volume CLEM, with node SUPPLY contributing an average of 0.14 I/O operations per second, node DEMAND contributing 1.46, and node NODE1 accounting for the remaining 0.42 I/O operations per second.
Average number of I/O requests waiting for service to the disk. The average queue size for each disk is followed by a breakdown of this value from contributing nodes, when applicable.
Average number of kilobytes per second transferred to or from the disk.
Average size of the I/O operations to the disk, in pages. The average size for each disk is followed by a breakdown of this value from contributing nodes, when applicable.
Name of the node that uses the various disk volumes. For rows in which no source node name appears, the data refers to the cluster-wide activity on the disk volume.
Name of the node that services I/O requests. An asterisk denotes that more than 1 server existed for this disk.
Percentage of time that I/O requests are outstanding to the volume for each node utilizing the disk volume.
Percentage of total I/O activity devoted to read operations. This value for each disk is followed by the percentage of read operations to the disk from contributing nodes, when applicable.
Percentage of total I/O activity that were split I/O operations. This value for each disk is followed by the percentage of read operations to the disk from contributing nodes, when applicable.
Disk volume type. In this example, all but one of the disks are RA81s.
Count of the Performance Manager records containing data for a disk volume during the interval.