|
|
|
An outage is considered to have occurred for a given prefix over a single link whenever the time interval between measurements through the link exceeds the larger of the following values:
Consider the following two cases, with minimum-silence-time and multiplier at their default values (10,000 and 10 respectively):
| Links |
Case 1 |
Case 2 |
|---|---|---|
Three links are all receiving measurements from prefix_a. In both cases, provider_2 stops receiving measurements.
In case 1, the CNA system determines there is a service outage after 10 seconds. The multiplier value times the maximum average interval across all links is 5,000 (10 times provider_3’s average of 500 ms), but the minimum-silence-time value is 10,000 milliseconds.
In case 2, the outage is determined after 50 seconds. The minimum-silence-time value remains 10,000 milliseconds, but the multiplier value times the maximum average interval is 50,000 ms (10 times provider_3’s average of 5,000 ms).
Under certain circumstances--when the currently asserted route experiences an outage blocking all UTT transactions to the affected prefix, for example--no measurements will be received across any of your links. When this happens, a link-comparison algorithm will not work.
The outage-detection total-silence command provides min, max, and multiplier parameters to be used to detect an outage across all links.
The CNA system calculates an initial interval as the product of the multiplier value and the largest of the most recently received IMTs for the prefix across each of your links.
If this initial interval is less than the user-configured min parameter, it is reset equal to min.
If the calculated interval is greater than the user-configured max parameter, it is reset equal to max.
An outage is declared for the prefix across all of your links if no measurements are received on any link for the period of time specified by this new interval.
A new interval is calculated every time a measurement is received, on any link, for the prefix.
|
|
|