Kofax Communication Server 9.1 - Manual (Page 48)

Environment Guide Version 3.00.04
48 © Copyright Kofax, Inc. All information is subject to change without notice.
The measurement cycle is by default set to 60 seconds, and can be configured in the following registry
value
(but it cannot be set lower than to 60 seconds, each value lower than 60 seconds would be set back to 60
seconds !)
HKLM\Software\TOPCALL\TCOSS\PerfCounterCheckCycle
Following table provides an overview of Key Performance Indicators that are being checked, their default
threshold limits and corresponding event log IDs:
KPI thresholds (in ms)
Configuration
(HKLM\SW\TOPCALL\TCOSS\...) EventLogID
Warning Information
KPI
(1)
Default thresholds
Local Disk Write
Latency [ms] Avg. 20 DiskLocalAvgAccessTime 16054 16060
Peak 2000 DiskLocalPeakAccessTime - 16055
Remote Disk Write
Latency [ms] Avg. 25 DiskRemoteAvgAccessTime 16056 16061
Peak 2000 DiskRemotePeakAccessTime - 16057
RTT
prim-sec
Avg. 10 DiskAvgNetworkDelay 16058 16062
Peak 500 DiskPeakNetworkDelay - 16059
RTT
ls1
RTT
FoIP
Avg. 300 NodeAvgRoundTripTime 16063 16065
Peak 1000 NodePeakRoundTripTime - 16064
RTT
storage-media
Avg. 10 MediaServerAvgRoundTripTime 16066 16068
Peak 500 MediaServerPeakRoundTripTime - 16067
(1) Please refer to the chapter Resource Estimation and Key Performance Indicators for detailed KPI
description
Each event log entry saves following additional information on the counter threshold violation:
1. Counter name and its value that violated its limit (average or peak)
2. In the case of average counter violation, the corresponding peak value of the same counter (and vice
versa)
3. Information in percent how many values exceeded the average and peak limits
4. Information on the limit values for both average and peak thresholds
5. The duration of the measurement cycle (default 60 seconds)
Note that it can happen during an idle period of TCOSS (e.g. no faxes being currently active) that the
measurement cycle indicated would be much longer that the configured value (e.g. 7000 seconds). This is
cause by the fact that counter verification is done only during regular TCOSS (disk) activity.
Example of the Warning event log written in the case of average counter threshold violation:
Avg.local disk write time 30 ms exceeded avg.limit, peak was 360 ms, 0 perc. values
exceeded peak limit (1000 ms), 53 perc. values exceeded avg.limit (25 ms) during
last 60 sec
Example of the Information event log written in the case above average counter does not exceed its limit
anymore:
Avg.local disk write time 19 ms is within avg.limit again, peak was 360 ms, 0 perc.
values exceeded peak limit (1000 ms), 53 perc. values exceeded avg.limit (25 ms)
during last 60 sec