In recent years, the concepts of Artificial Intelligence (AI) and Machine Learning (ML)...
3.12.4 Core configuration
All Core parameters affecting the decision-making algorithms can be modified on the next tab (“Configuration→Core”).
Core #
The following configuration parameters can be adjusted:
- Max IPv4 improvements (core.improvements.max, core.improvements.max_ipv6)
- Standard reprobing period (core.improvements.ttl.retry_probe) – the time period after which a specific improvement is being scheduled for re-probing.
- VIP reprobing period (core.vip.interval.probe) – defines the VIP prefixes probing interval, in seconds.
- Minimal probe lifetime (core.probes.ttl.min) – Ordinary probing will not be performed for a specific prefix if its probe age is lower than this value.
- Allowed latency worsening (core.cost.worst_ms)
- Exploring queue slots (core.eventqueuelimit)
- Relevant loss for loss-based improvements (core.performance.loss_pct) – a prefix will be improved based on a loss cause only if the packet loss can be decreased by this value (in %)
- Relevant RTT difference (core.performance.rtt.diff_ms)
- Relevant RTT difference (%) (core.performance.rtt.diff_pct)
- Maximum probe lifetime (core.probes.ttl.max)
Outage Detection #
Outage detection algorithm can be enabled and configured on the same page. Outage detection algorithm can be enabled or disabled using the toggle buttons at the top right.
Overusage #
Throttling with Flowspec can be configured on the same page.
The following configuration parameters can be adjusted:
- Overusage interval (core.overusage.check_interval) sets the frequency in seconds of checking for excessive bandwidth use.
- Overusage rule retention (core.overusage.hold_timer) sets how much time a rule is kept after bandwidth use returns to normal.
- Prefix BW average time (core.overusage.out.average.period) sets the number of hours used to determine the average bandwidth use of a prefix.
- Prefix relevant BW (core.overusage.out.average.relevant_min) sets the relevant bandwidth use in Mbps by a prefix before considering any rules for it.
- Overusage throttle multiplier (core.overusage.out.threshold.throttle) sets the multiplier to apply to average prefix bandwidth use when setting a rule.
- Overusage threshold multiplier (core.overusage.out.threshold.trigger) sets the threshold multiplier used to determine excessive bandwidth use.
Prefix relevant BW and Overusage threshold multiplier parameters control the number of Flowspec rules that will be generated automatically. Adjust these values to control the number of Throttling Flowspec policies.
Circuit Issues Detection #
Circuit Issues Detection parameters are listed and can be adjusted:
The following configuration parameters can be adjusted:
- Delta loss to shutdown (core.circuit.high_loss_diff) sets threshold to initiate shutting down a provider with circuit issues as a difference between examined provider’s average loss and overall average loss during the given time horizon. Shutdown is attempted only for providers marked accordingly.
- Delta loss to warn (core.circuit.warn_loss_diff) sets threshold to raise a warning regarding issues with a provider as difference between examined provider’s average loss and overall average loss during the given time horizon. Usually this threshold is significantly lower than the shutdown threshold.
- Delta loss to restore (core.circuit.recover_loss_diff) sets low loss level threshold when IRP will restore full functionality over provider that had circuit issues.
- Issues time horizon (core.circuit.hist_interval) sets how many minutes in the past IRP looks for probes with loss over both analyzed provider and all the other providers on the network.
- Withdraw improvements on warn (core.circuit.withdraw_on_warn) instructs IRP to withdraw outbound improvements over provider with circuit issues when warning threshold is reached. By default improvements are withdrawn only when shutdown threshold is exceeded.
- Restore after (core.circuit.recover_hold_time) sets the interval in seconds after which IRP should re-evaluate a provider’s circuit loss and attempt restoring it to normal function. This will be performed only for providers that are configured to attempt to restore after shutdown.
- Restore interval (core.circuit.recover_monitored_intervals) sets the interval in minutes during which IRP will periodically re-evaluate a provider’s circuit loss and attempt restoring it to normal function. This will be performed only for providers that are configured to attempt to restore after shutdown.
The configuration parameters above are applied for all providers even though for individual providers different level of control can be set starting from disabling and ending with attempting both to automatically shutdown and later restore connectivity with it. See also Providers configuration