2 checking disk io latency, Using tcdisktest tool – Kofax Communication Server 9.1.1 User Manual

Page 41

Advertising
background image

Environment Guide

Version 3.00.04

© Copyright Kofax, Inc. All information is subject to change without notice.

41

Memory active is the amount of memory that the VM is actively using, based on the current VM activity,

application

s running there etc. (on the other hand, Memory consumed is the memory the VM is “occupying”

on the ESX server, despite of the current VM activity).

As a rule of the thumb it can be said that memory active should always be lower than memory consumed

for th

e proper VM‟s operation without a lot of RAM swapping.

Further, in the case of real-time applications, check whether the Memory active values are permanently

lower than values estimated for RAM in the resource estimation tables.

In the screenshot above,

up to about 10:00 o‟clock the VM didn‟t have any fix RAM reservation, and it can

be seen that the ESX server grants dynamically a little more memory than really needed (Memory active

counter).

After 10:00, the VM has been assigned about 512 MByte of RAM, and it can been seen that the Memory

granted counter remains constantly on this level despite of the amount of memory actively being used.

5.4.2

Checking Disk IO Latency

The disk bandwidth is the most critical resource as the physical disk is typically implemented as some kind

of network disk arrays (SAN) and its bandwidth will be shared by all VM running on the particular ESX

server.

If the administrators properly calculate disk bandwidth requirements of all involved applications, there

should be no problem but often peak requests from other applications may also influence the available disk

bandwidth of the KCS platform applications. While this is not a real problem for non real-time applications

(may lead to increased end user response times for TC/Web etc.), it may cause substantial problems

especially on the TCOSS server like:

1. Interrupted faxes being sent or received

2. Reboots and resynchronization of the secondary master in the case of tandem server operation

Therefore, it is a good idea to verify the disk bandwidth availability even prior to but also continuously during

the TCOSS operation to see whether enough disk bandwidth is available.

Using TcDiskTest Tool

TcDiskTest is a simple command-line utility that performs disk write operations in the same manner TCOSS

would do. It is easily possible to test the suitability of a particular virtual machine/virtualized environment for

the desired TCOSS operation. During the first run the tool creates own test file (named tcDisktest.bin) in the

root directory of the disk/partition to be tested, and then performs random write operations of a data block of

4-kBytes (configurable) to this file and exports three performance indicators as Windows performance

monitor counters:

1. Disk ms/Write (Current disk write access latency in ms)

2. Avg.Disk ms/Write (Average disk write access latency in ms)

3. Peak Disk ms/Write (Peak disk write access latency in ms, since last performance monitor sample)

Installation

Advertising