PG_COLUMN_SIZE

Return the number of bytes used to store a value in its internal representation.

Category: miscReturns: INTDialect: PostgreSql

Syntax

PG_COLUMN_SIZE(expr)

Description

## Overview Returns the number of bytes used to store the given value in DeltaForge's internal representation. Use this function to estimate table size, identify unusually large values, or build storage-efficiency reports. The measurement reflects the internal encoding, which may differ from the on-disk size after compression. ## Behavior - Returns NULL if the input is NULL. - Returns an INT byte count. - For variable-length types (strings, arrays), the result depends on the actual value length. - Deterministic with respect to the input value. - Side effect free. ## Compatibility - PG-compat alias for `pg_column_size` function.

Parameters

Name	Type	Description
`expr`		Specifies the expression whose internal storage size is measured.

Examples

-- Size of a small integer
SELECT PG_COLUMN_SIZE(42);

-- Size of a short string
SELECT PG_COLUMN_SIZE('hello world');

-- Compare sizes across numeric types
SELECT PG_COLUMN_SIZE(1::SMALLINT) AS small, PG_COLUMN_SIZE(1::BIGINT) AS big;

-- Per-column size statistics in a report
SELECT PG_COLUMN_SIZE(name) AS name_bytes, PG_COLUMN_SIZE(description) AS desc_bytes
FROM retail.catalog.products;

-- NULL input returns NULL
SELECT PG_COLUMN_SIZE(NULL);  -- NULL

Pitfalls

PG_COLUMN_SIZE returns the internal-representation size, not the on-disk size after compression. Parquet or Delta compression can drastically reduce the footprint.
Sizes for variable-length types depend on the value, not the column type. Averaging PG_COLUMN_SIZE over a sample gives a better estimate than measuring a single row.
The function does not account for per-row storage overhead (headers, null bitmaps). Summing PG_COLUMN_SIZE across columns under-counts the true row size.
Results for TOAST-candidate types in the PG-compat model reflect the uncompressed inline size, which may not match the actual storage footprint.