Metrics & alert rules

Metrics

Charts render recent activity at full resolution and downsample automatically for wider ranges using 5-minute rollups, so dashboards stay fast even at thousands of databases. Pick a period (1h–7d) per connection.

Health score

The Health Score tab runs every assessment at once on PostgreSQL, MySQL/MariaDB, SQL Server, Oracle, and IBM Db2 — configuration audit, maintenance advisor, lock contention, wait mix, and replication/HA — and synthesizes them into one weighted 0–100 score and A–F grade. A category breakdown (Security, Reliability, Performance, Maintenance, Availability, Observability) shows where points were lost, and a ranked "top issues" list links each deduction to the underlying finding — the fastest way to triage a database and decide what to fix first.

Two alerting modes

Create a rule with a metric, severity, and a detection mode:

• Static threshold — fires when the metric crosses a fixed value you set.
• Anomaly — fires when the metric deviates sharply (z-score) from its recent baseline, so there is no fixed number to tune.
• Acknowledge or resolve events from the feed; a cooldown prevents alert spam.
• The composite health score is itself an alertable metric — e.g. page when health_score drops below 70 — closing the loop from assessment to proactive alerting.
• Capacity is alertable too: database size and growth-rate (bytes/day) are sampled hourly, so you can page before a disk fills — alert when a database exceeds a size ceiling or grows faster than a set rate.

Storage insights

The Storage tab breaks capacity down by object on PostgreSQL, MySQL/MariaDB, SQL Server, Oracle, and IBM Db2 — the total database size plus the largest tables, each split into table (heap) vs index bytes with an estimated row count and size-share bar (from each engine's own catalog — e.g. pg_total_relation_size, information_schema, SYSIBMADM.ADMINTABINFO). An index footprint that exceeds its table is flagged, pairing naturally with the Maintenance Advisor to find unused indexes. The platform also samples sizes continuously (about hourly) to chart database growth over 7d/30d/90d and rank the fastest-growing tables with a 30-day forecast — so you know which object will fill your disk, and roughly when.