Analyze Performance › Analysis Reports › Interpret the Analysis Reports › Cluster Rule Analysis Report

Cluster Rule Analysis Report

The following example shows a Cluster Rule analysis report:

Full Analysis                        CLUSTER                            Page   6 
							PA Vx.x
	Saturday 01-OCT-1996 00:00 to 23:00
CONCLUSION 1.
							{L0050}
Queues are  forming  on  heavily  used  disks.   Longer delays will be experienced when longer queues
form.

Suggested Remedies:

1. 	If the disk is fragmented (use the command $ ADVISE COLLECT REPORT DISK_SPACE devicename in
order to check on this), backup and  restore  the disk using the BACKUP utility without the
/PHYSICAL qualifier.
2. 	If possible, assign all new work during  the  noted times to other disk volumes.
3. 	If possible, attempt to lower the future  usage  of the noted disk volumes.

Volume name(s)      Number of Samples 
--------------      ----------------- 
PROBLEM_DISK                   4

CONDITIONS

1.	DISK_QUEUE_AT_SERVER .GE. 1.00 .OR. ( MAXIMUM_DISK_QUEUE .GE. 1.00 AND.DISK_IS_SERVED .EQ.
0.00 )
2.	CW_DISK_IO_RATE .GE. DISK_IO_RATE_THRESHOLD .OR. CW_DISK_THRUPUT_RATE .GE.
DISK_THRUPUT_RATE_THRESHOLD 
3.	OCCURRENCES .GE. 1 

EVIDENCE
Disk Volume      Srv Node               Time of Occ    Shdw Rbld
    \                     Avg I/O    Avg   IO Sz
    \Src Node             per    sec  Que  Pages    Hottest File
-------------------------------------------------------------------------------- 
  1                     2                	           3                    4
PROBLEM_DISK MARCUS          25-JUN 00:06:00          *          
      5                                     6         7            8            9
      SUPPLY           	25.80  2.17      6.0 [PSCP]PSCP010.B;9 
      DEMAND           	32.80  1.64    16.0 [PSALL]BIGFILE.DAT 
PROBLEM_DISK MARCUS          25-JUN 00:12:00 
      SUPPLY           	20.59  1.60      5.6 [PSDC]PSDC010.B;21 
      DEMAND          	16.55  1.73      5.6 [PSDC]PSDC010.B;21 
PROBLEM_DISK MARCUS          25-JUN 00:16:00 
      SUPPLY           	28.59  1.64      5.6 [PSDC]PSDC010.B;21 
      DEMAND           	31.33  1.87      6.3 [PSPA]PSPA010.B;14 
PROBLEM_DISK MARCUS          25-JUN 00:26:00 
      SUPPLY           	24.59  1.55      5.6 [PSDC]PSDC010.B;21 
      DEMAND           	29.58  1.97      6.2 [SMITH.WORK.PSCP]PSCP$MAIN.EXE;67

The following statements are keyed to the previous Cluster Rule analysis report example:

1. Name of the disk experiencing heavy use.
2. Name of the node that actually services requests.
3. Beginning time of the interval in which the condition occurred.
4. An asterisk indicates that a disk is a shadow set and the disk underwent a COPY operation.
5. Name of the node in the cluster that shares the heavily used disk.
6. Average number of operations per second to the volume, during the given interval, by the contributing node.
7. Average size of the queue during the interval exampled, measured by the number of requests.
8. Average size, in pages, of all I/O requests during the interval.
9. When hot file data exists, the hottest file (highest I/O rate)is listed.

In the previous Cluster Rule analysis report example, the queue length on volume PROBLEM_DISK exceeds the value of 1.0 on four occasions. During those four occasions, the total operations per second for each interval exceeded the device threshold.