Kofax Communication Server 9.1.1 - Manual (Page 44)

Environment Guide Version 3.00.04
44 © Copyright Kofax, Inc. All information is subject to change without notice.
Setting of the Test Cycle Parameter
With a higher test cycle value (e.g. 1000ms) the TcDiskTest tool would make fewer disk IOs than a fully
utilized TCOSS server, but anyway it may still detect disk bottleneck in the virtual environment if the disk
resource is full overloaded.
For example, starting TcDiskTest with the command line
TcDiskTest -p1000 -t360000 -dd
would produce around 4kByte/s transfer data towards the disk, which is much more less than TCOSS
would.
On the other, it could also make sense to simulate the same disk transfer capacity similar with TCOSS in
order to see, whether the virtualized environment would provide this capacity also for the longer time.
In this case proceed as follows:
Compute the test cycle based on the disk IOPS value given in the chapter Resource estimation and Key
performance Indicators. For example, for the 60-channels TCOSS fax server IOPS=36,
so the test cycle = 1000ms/36 = 27ms
To generate disk load similar to what TCOSS would do for the duration of 100 hours, start the TcDiskTest
with the following parameters:
TcDiskTest -p27 –t360000 –dd
Setting the test cycle parameter would produce higher disk load, but would better test the suitability of
particular ESX environment for the TCOSS operation.
Using TCOSS Internal Performance Counters
Following TCOSS disk Windows performance counters provide a very good means to measure the Key
performance Indicators of the TCOSS Server.
(“local disk” counters are always available, “remote disk” counters are only available on the primary master
in the tandem operation):
Avg.Local Disk ms/Write (Average local disk write access latency in ms)
Peak Local Disk ms/Write (Peak local disk write access latency in ms)
Avg.Remote Disk ms/Write (Average remote disk write access latency in ms, Tandem Server)
Peak Remote Disk ms/Write (Peak remote disk write access latency in ms, Tandem Server)
Avg.remote Disk Network delay ms (Average network latency in ms, Tandem Server)
Peak. remote Disk Network delay ms (Peak network latency in ms, Tandem Server)
Average counters (“Avg.”) indicate the average of the particular value between two subsequent
performance monitor measures.
Peak counters (“Peak”) indicate the peak value between two subsequent performance monitor measures
The Avg. and Peak Remote Disk ms/Write counters indicate the whole duration of the remote disk write
access (network latency + write disk operation).
The Avg. and Peak remote Disk Network delay ms counters indicate the network portion of the remote disk
write operation (network latency).