Dashboard

The landing page — an operational overview of what your data is doing, not object counts (those collapse to a single Schema: line: tables · views · procedures · triggers · indexes · functions · event triggers, each a link). Each card loads independently and shows a small “Couldn’t load…” note on failure rather than blanking the page.

• Health row (shown to users with data.ingest): Sync health (24h) — imports succeeded / failed / in-flight (error accent if any failed); Data freshness — when the most recent sync landed and the stalest enabled entity (its last run, or “never run”); Rows moved (24h) — rows inserted / updated / deleted across all syncs; Attention needed — a rollup of connection warnings, recent Data Quality mismatches, webhook dead-letters, and agents with an update available (“All clear” when empty).
• Detail row (shown with data.ingest, tables.view, or dab.view): Recent failures — latest failed runs with entity/time/error, linking to Import; Storage — the tenant DB’s data/log file fill (used / size / percent, “of max” when a max is set) plus the largest tables by reserved space (row count · MB), read from SQL Server file/partition stats (no full scans); Usage — request volume over 24h and 7d split into API calls and DAB (incl. MCP) calls — MCP tool traffic rides the same DAB containers and DAB metrics don’t tag MCP separately, so the DAB figure inherently includes MCP (shows “Metrics unavailable” if the source can’t be reached, not a misleading zero).
• Quick Actions & Recent Activity — permission-respecting shortcuts (create objects, manage users/permissions, Access Management, Browse Data, Settings) plus recent audit activity (or a workspace summary without audit-view access).

A tenant that hasn’t configured any connectors sees empty health cards — expected, not an error.