The following example shows disk statistics in By Node format:
Tabular Report CLUSTER Page 1 PA Vx.x Tuesday 26-JAN-2006 09:00 to 09:20 ************************ Final Statistics *********************** ************************ Data Analyzed: from 26-JAN-2006 09:00:00.00 to 26-JAN-2006 09:20:00.00 *********************** +----------------------------------- BY NODE Disk Statistics -------------------------+ ! ! ! Work Resp ! ! Avail Paging Swping Rate Time Queue Space ! ! % % % (/s) (ms) Length Used % ! ! ------ ------ ------ ------ ------ ------ ------ ! ! DSA0 ! ! LATOUR 0.0 0.0 0.0 0.0 0 0.0 96 ! ! YQUEM 0.0 0.0 0.0 0.0 0 0.0 96 ! ! GALLO 0.0 0.0 0.0 0.0 0 0.0 96 ! ! DSA1 ! ! LATOUR 0.0 0.0 0.0 0.0 0 0.0 81 ! ! YQUEM 0.0 0.0 0.0 0.0 0 0.0 81 ! ! GALLO 0.0 0.0 0.0 0.0 0 0.0 81 ! ! DSA10 ! ! LATOUR 1.9 0.0 0.0 0.6 37 0.0 61 ! ! YQUEM 3.7 0.0 0.0 1.0 40 0.0 61 ! ! GALLO 1.3 0.0 0.0 0.4 36 0.0 61 ! ! DSA111 ! ! LATOUR 16.9 25.9 0.0 7.1 29 0.2 99 ! ! YQUEM 9.3 27.9 0.0 3.1 36 0.1 99 ! ! GALLO 7.7 26.3 0.0 2.2 39 0.1 99 ! ! . . . . . . . . ! ! . . . . . . . . ! ! . . . . . . . . ! +-------------------------------------------------------------------------------------+ User Command: ADVISE PERF REPORT TABULAR=(BYNODE)/BEG=26-JAN-2006 09:00:00.00 - /END=26-JAN-2006 09:20:00.00/OUT=BYCLU.TAB/SECTION=DISK
In the By Node format for disk statistics, the leftmost column of the tabular report contains the name of each disk, followed by the name of each node from which the disk was accessed. Following each node name is the processor type; and if the node hosted the disk, the word host.
Each disk, with node names, begins a line of disk statistics similar to the By Cluster format. These statistics show the contribution of each individual node to the cluster-wide activity of a specific disk. The way the Performance software determines a node's contribution to the cluster-wide activity of a disk depends upon whether the node has direct or remote access to the disk. For nodes with remote access, the I/O statistic for each node is that node's contribution to the total cluster I/O load.
In the case of a configuration with one node with direct access to the disk and any number of nodes with remote access, I/O statistics for the node with direct access includes I/Os from any nodes with remote access to the disk. Therefore, to compute the direct access node's contribution, the software subtracts the number of I/Os of the nodes with remote access from the I/O information of the node with direct access.
In the case of two nodes with direct access to a disk with any number of nodes with indirect access, both direct nodes' I/O statistics include I/O from any remote nodes. Therefore, it is not possible to compute the By Node contribution to the cluster total of the nodes with direct access to the disk. The By Node contribution of the nodes that access the disk remotely is each node's I/O statistics.
For example, the software would calculate By Node rates for disks DU1 and DB2 in the example configuration as follows. For disk DU1 with the HSC, all I/Os come to this disk through host (direct access) Nodes A and B. The I/O information for these nodes includes I/O from the served (remote access) Node C. There is no way to distinguish whether Node C's I/Os were processed by Node A or Node B. Therefore, calculating the By Node contribution for Nodes A and B, which directly access the disk, is not possible. The By Node contribution for Node C, which has remote access to Disk DU1, is the I/O statistic for Node C.
For disk DB2, Node C is the host (direct access) node; all I/Os come to this disk through Node C. Therefore, to calculate Node C's By Node contribution, subtract from Node C's I/O information the I/O information from Nodes A and B, which have remote access to DB2. The By Node contribution of Nodes A and B, is each node's I/O statistics.
|
Copyright © 2008 CA.
All rights reserved.
|
|