The current SolarWinds Platform alerting engine relies heavily on a flat time-delay rule: "Condition must exist for more than X minutes." This rigid, time-based approach is deeply flawed for dynamic infrastructures where different objects have varying polling intervals.
We need a native feature to evaluate raw polling history ratios directly inside the standard Alert Wizard. Administrators must have the ability to define a specific number of polls, where an alert triggers if a condition is met across a fraction of those polls (e.g., "X out of Y polls").
Real-World Scenario Example:
- An alert check interval is set to run every 2 minutes.
- The administrator configures a metric threshold with a window of 4 polls.
- If 3 out of those 4 recent polls indicate that latency is breached, the alert triggers.
- This allows isolated transient spikes to be ignored while capturing sustained degradation, without relying on a flat time clock.
Why the Current Architecture Fails:
- Poll Interval Discrepancies: If different nodes or interfaces poll at different intervals (e.g., some at 2 minutes, some at 10 minutes), a global "X minutes" rule cannot evaluate them fairly. A 10-minute delay represents 5 polling cycles for one device, but only 1 cycle for another.
- Flapping Protection is Fragmented: While SAM component templates allow "X out of Y" logic natively, this critical capability is entirely missing from the core alerting framework used by NPM, VMAN, SRM, and other core platform modules.
Requested Solution:
Please extend a native condition modifier into the standard SolarWinds Alert Wizard dropdown menus. : "Trigger if condition is met in [X] out of the last [Y] polling cycles." This will provide uniform, granular, and poll-accurate notification control across the entire SolarWinds platform ecosystem.
