Health score & assessments
One composite grade per database, backed by a suite of deep, read-only assessments.
What the assessment suite is
Beyond raw metrics, the platform runs a set of read-only assessments against each database and synthesizes them into a single, comparable health grade. Everything is read straight from the engine's own catalogs and dynamic views — nothing is changed, and no agent-side state is required. The assessments span five SQL engine families (PostgreSQL, MySQL/MariaDB, SQL Server, Oracle, IBM Db2), with NoSQL engines shown as unsupported where an assessment has no equivalent.
Composite health score
The Health Score tab computes one weighted 0–100 grade (A–F) per database by synthesizing configuration, reliability, performance, maintenance, availability, and observability signals. A category breakdown shows where points were lost, and a ranked "top issues" list links each deduction back to the underlying finding. A fleet leaderboard ranks every database worst-first, so you always know what to fix next — and health_score is itself an alertable metric (e.g. page when it drops below 70).
Activity & wait events
The Activity view samples active sessions live (every few seconds) and groups them by what they are waiting on — CPU, locks, I/O, and engine-specific wait classes — with a live-vs-historical Active Session History chart. A "Top SQL by wait" panel attributes the waiting to the exact statements, so you go from "the database is lock-bound" to the specific query responsible. From there, one click runs an AI analysis of that query.
Storage insights & capacity
The Storage tab breaks capacity down per object — the largest tables and their indexes with total database size — and continuously samples sizes to chart growth over 7d/30d/90d and project which table will fill the disk, and roughly when. Database size and growth-rate are alertable metrics, so you can page before a disk fills.
Configuration audit & CIS mapping
The Configuration Audit reads the live server configuration and flags risky or suboptimal settings (TLS, durability, autovacuum, timeouts, weak password hashing, logging, dangerous features) with the current value, the recommended value, and why it matters. Every finding is mapped to its CIS Benchmark domain, and an "open gaps by domain" rollup turns the audit into a compliance-posture view. A fleet-wide view aggregates those CIS gaps across every database in the workspace and ranks connections by most findings first, so a compliance owner sees the whole estate's posture at a glance and drills into any one.
Advisors & lock contention
The Index Advisor mines captured slow queries for missing-index patterns (ready-to-review CREATE INDEX DDL); the Maintenance Advisor reads each engine's statistics catalogs to surface bloat, unused indexes, stale statistics, fragmentation, and tables missing optimizer statistics. The Session Manager's lock-contention panel shows every blocked session paired with the session blocking it — who waits on whom — and lets you terminate the blocker to release the chain, all audited.
Snapshots & diff — did my change help?
Capture a point-in-time snapshot of a database's assessment (with an optional note like "before autovacuum tuning"), then compare it against a fresh live assessment. The diff shows the score delta and grade change plus every issue grouped into regressed/new, resolved, and severity-changed — so after a tuning change, index build, or config edit you can prove exactly what got better or worse instead of eyeballing two scores.
On-prem & agent-brokered databases
Every assessment above works identically whether a connection is direct or reached through the mTLS on-prem agent — including the composite health score and the fleet leaderboard. The agent gathers the underlying signals inside your network and the platform scores them, so the same read-only operations reach databases that never leave your network.
Try it yourself
Open the console and put this into practice.