Version

Argent Guardian 8.0A-0701

Date

3 Apr 2007

Summary

Argent Guardian Ping Rules Can Erroneously Return 0ms Response Time

Technical Background

An Argent Guardian ping rule can be configured to measure latency for the ping response, as follows:

This should save the ping latency as Argent Predictor data, keyed to Service Level Agreement (object) and System Response Time (counter) against the device of interest (instance).

Across reasonably fast links, values of 0ms are often reported.

For example, the above ping rule was run across a >2000km WAN link.

The response times recorded in the Argent Predictor database are retrieved with the following SQL query:

SELECT COUNT(*) as count, metric

FROM AG_PRD_SLAVE

WHERE (master_key = 714) and (snap_time > ’03/04/2007′)

GROUP BY metric

ORDER BY metric

count	metric
250700	0
8310	15
13706	16
751	31
264	32
9	46
46	47

Across this WAN link (>2000km), a 0ms round trip time is not physically possible.

It appears as though the value reported for ping latency is granular to just less than 16ms, and frequently rounded down to 0ms.

This limitation is not present in the ping.exe included with Win200x, in addition to other monitoring tools.

Resolution

There is no known workaround to this issue.

An issue report has been logged and it will be addressed in a future release.