Skip to main content

Multi-metric tables

A tag + field model for multi-metric time-series data. A multi-metric ChronoTable indexes its tag columns automatically and exposes cardinality controls so you can catch label explosion early.

create_metric_table(p_table_name, p_tag_columns, p_field_columns, p_chunk_interval, p_schema_name)​

Creates a ChronoTable optimized for multi-metric data with automatic indexing on tag columns.

ParameterTypeDefaultDescription
p_table_nameTEXT—Table name
p_tag_columnsTEXT[]—Tag columns (indexed TEXT, low cardinality)
p_field_columnsTEXT[]—Field columns (DOUBLE PRECISION, high cardinality)
p_chunk_intervalINTERVAL'1 day'Partition interval
p_schema_nameTEXT'public'Schema

Returns: INT — chronotable_id

SELECT lakets.create_metric_table(
    'system_metrics',
    ARRAY['host', 'datacenter', 'service'],   -- tags
    ARRAY['cpu_pct', 'mem_pct', 'disk_io'],   -- fields
    '1 day'
);

cardinality_stats(p_table_name, p_schema_name)​

Distinct value counts per tag column — essential for catching series explosion.

Returns: TABLE — column_name, distinct_values, total_rows, pct_of_rows

cardinality_check(p_table_name, p_max_series, p_schema_name)​

Warns if combined cardinality across all tag columns exceeds a threshold.

ParameterTypeDefaultDescription
p_table_nameTEXT—Metric table name
p_max_seriesBIGINT100000Maximum allowed series count
p_schema_nameTEXT'public'Schema

Returns: TABLE — status ('OK', 'WARNING' above 80% of the limit, or 'CRITICAL' above the limit), combined_cardinality, max_allowed, tag_columns

SELECT * FROM lakets.cardinality_check('system_metrics');
-- OK | 2500 | 100000 | host,datacenter,service