Advanced Network Performance Monitor Settings enable you to decide how to deal with the following actions:
Default values are selected for all of these options that work for most networks. Ensure you understand the implications of changing any of the values you decide to modify.
When Toolset Network Performance Monitor starts, current data is unavailable for your network. Toolset Network Performance Monitor creates a baseline. All Resources are polled immediately, and when the first poll is complete, the network is polled again. These two result sets are used to create statistics providing an immediate view of network performance.
Baseline calculation requires a great deal of data gathering and data computation that can affect the performance of the Toolset Network Performance Monitor host and polled router. If you do not need statistics immediately upon starting Toolset Network Performance Monitor, select Disable Baseline Calculation at Startup on the Baseline Calculation tab.
Toolset Network Performance Monitor provides two methods for calculating device availability. The default method uses the status of the node, up or down. As long as the device responds to a ping within the warning interval, Toolset Network Performance Monitor considers the node up. For more information about the node warning interval, see Node warning interval
The other method offered bases node availability on packet loss percentage. Unless you need packet loss percentage based availability calculations, we recommend that you retain the default method for availability calculation.
Devices drop packets and fail to respond to polling for many reasons. When a device fails to respond, Toolset Network Performance Monitor changes the status from Up to Warning. The node warning interval can specify how long a device can remain in a Warning state before it is considered unreachable or Down. While a node is in the Warning state, the service performs fast polling, pinging the device every 3 minutes, to detect any change in node status.
If you see erroneous events or receive false alerts for down nodes, it is possible that the application is detecting intermittent packet loss within the network. If you do not want to diagnose the cause of the network flutter, consider setting the node warning interval to a higher value.
Depending on the monitored device enter and your Toolset Network Performance Monitor settings, statistical information may be derived from 32-bit counters. These counters offer a maximum value of 2 to the 32nd power (4,294,967,296). You can decide how Toolset Network Performance Monitor compensates when a 32-bit counter reaches maximum capacity and rolls over to zero. The default method detects a counter rollover and ignores the zeroed value. Instead, a new sample is scheduled in 20 seconds. This method can cause slightly skewed total bytes transferred numbers.
The more advanced method, which produces skewed numbers when manual counter resets occur, detects a counter rollover and uses the following calculation to correct for the rollover to zero: currentValue + (maximumCounterValue - lastCounterValue). If manual counter resets are rare, this method produces the most accurate results.
Toolset Network Performance Monitor can and does gather statistics from the 64-bit counters on network devices, if available. These counters have a capacity of 2 to the 64th power (18,446,744,073,709,551,616). While Toolset Network Performance Monitor fully supports the use of 64-bit counters, the implementation of these high capacity counters by some hardware vendors exhibit erratic behavior. If you notice peculiar results when using these counters, disable the use of 64-bit counters for the device and contact the hardware manufacturer.
You can enable and disable 64-bit counter use on the Node Details window. See View node data and modify node properties
Toolset Network Performance Monitor can save periodic XML files of gathered data. This XML Snapshot can be used by other programs, most commonly Web applications, to display information about selected devices or interfaces.
Generating XML snapshots causes intensive CPU usage. Leave this option disabled unless you use the XML Snapshots.