In this chapter
This chapter describes the time synchronization statistics that can be obtained for all time synchronization modes for SmartNICs with Napatech Link-Capture™ Software.
This table shows the time synchronization statistics. The statistics are calculated on the differences between the time of the SmartNIC time-stamping clock and the time provided by a time synchronization event.
|Standard deviation||Standard deviation in ns of offset from reference
measured since last reset.
Note: The standard deviation computed in the time synchronization statistics may deviate up to 1 ns from the standard deviation that can be obtained from a sequence of sampled NT-TS values. The statistics are computed from 100 of the 50,000 time synchronization pulses received per second, while the NT-TS samples are taken once per second. Therefore the two results may be based upon different signals.
|Peak-to-peak jitter||Difference in ns between high and low offset measured since last reset .|
|Current skew||Last measured offset from reference in ns.|
|Lowest skew||Lowest offset from reference in ns measured since last reset.|
|Highest skew||Highest offset from reference in ns measured since last reset.|
|Mean skew||Calculated mean offset from reference in ns.|
|Time since reset||Number of seconds since last reset of information/statistics.|
|Loss of input reference signal||Number of times the reference signal has been lost since last reset.|
|Loss of synchronization||Number of times the synchronization has been lost since last reset.|
|Hard synchronization events||Number of hard synchronization events since last reset.|
The time synchronization statistics can be obtained by using the NT_InfoRead function in the information stream (see DN-0449) with the NT_INFO_CMD_READ_TIMESYNC_STAT command. The NtTsStatSupport_e enumeration shows to which degree the SmartNIC in question supports time synchronization statistics. The NtInfoTimeSyncStatistics_s structure contains the statistics.
The statistics can also be displayed using the monitoring tool (see DN-0449).
The statistics are reset one minute after a master signal has been detected for the first time. Thereafter the statistics are continuously updated as long as a skew against a reference master can be calculated. No reset is made in case of failover or fallback. The calculation of the skew will just be made against the new reference.
The time synchronization statistics can be reset from an application using the NT_ConfigWrite function (see DN-0449).