Skip to main content

Metrics Database

The metrics database stores raw cluster metrics, aggregated views, and customer profile data.

Tables

metrics_table (~830M rows)

The raw metrics store. Every metric data point from all BYOC clusters lands here.

ColumnTypeDescription
event_timeDATETIMETimestamp of the metric sample
account_idVARCHARReferences accounts.biz_id (short form)
cluster_idVARCHARReferences clusters.biz_id (UUID)
metric_nameVARCHARMetric identifier (see metrics_dict_v2)
valueDOUBLEMetric value
tagsJSONAdditional dimensions (node, warehouse, etc.)

metrics_dict_v2 (208 rows)

Metric dictionary covering 26 categories with units and human-readable descriptions. Use this to look up what a metric_name means, its expected unit, and which category it belongs to.

dim_customer_profile (~30 rows)

Customer dimension table for filtering and grouping.

Key ColumnsValues
tierA, B, S
stageVarious lifecycle stages
use_caseCustomer's primary StarRocks use case

Materialized Views

amv_hourly_snapshots_v1 (~33.7M rows)

Hourly aggregated snapshots of metrics. Pre-computes common aggregations to avoid scanning the full 830M-row metrics_table.

Views

metrics_hourly_view_v5 / metrics_hourly_view_v6

Enriched hourly metrics joined with metrics_dict_v2 metadata. v6 adds extra columns over v5.

v_customer_daily_360_v2 (~1,374 rows)

Executive daily summary per customer. The preferred version (over v1) for dashboards and reports. Aggregates key health indicators by customer by day.

v_customer_key_v2

Bridge view mapping biz_id to human-readable display names. Use this when you need to show customer names instead of IDs.