Snowflake External Volumes and Iceberg Integration: Complete Setup Guide

An External Volume is an object that defines the external cloud storage location where Snowflake stores data files (Parquet) and metadata for Apache Iceberg Tables. It points to S3, Azure Blob Storage, or Google Cloud Storage as the durable data backend, letting the Snowflake compute engine read and write Iceberg-format tables.

How External Volumes Relate to Iceberg Tables

Snowflake
  │
  ├─ CREATE ICEBERG TABLE ... EXTERNAL_VOLUME = 'my_vol'
  │    └─ Table metadata: Snowflake Catalog or External Catalog
  │
  └─ External Volume (my_vol)
       │
       └─ Cloud storage
            ├─ data/     ← Parquet data files
            └─ metadata/ ← Iceberg metadata (manifest, snapshot)

Creating an External Volume for AWS S3

CREATE OR REPLACE EXTERNAL VOLUME iceberg_s3_vol
  STORAGE_LOCATIONS = (
    (
      NAME = 'us-east-1-primary'
      STORAGE_BASE_URL = 's3://my-iceberg-bucket/warehouse/'
      STORAGE_PROVIDER = 'S3'
      STORAGE_AWS_ROLE_ARN = 'arn:aws:iam::123456789012:role/snowflake-iceberg-role'
    )
  );

-- Check External Volume details to configure IAM on the AWS side
DESCRIBE EXTERNAL VOLUME iceberg_s3_vol;
-- → STORAGE_AWS_IAM_USER_ARN
-- → STORAGE_AWS_EXTERNAL_ID
-- Add these to the trust policy of the AWS IAM role

Creating an External Volume for Azure Blob Storage

CREATE OR REPLACE EXTERNAL VOLUME iceberg_azure_vol
  STORAGE_LOCATIONS = (
    (
      NAME = 'azure-east-primary'
      STORAGE_BASE_URL = 'azure://myaccount.blob.core.windows.net/iceberg-container/warehouse/'
      STORAGE_PROVIDER = 'AZURE'
      AZURE_TENANT_ID = 'a1b2c3d4-e5f6-7890-abcd-ef1234567890'
    )
  );

-- After DESCRIBE, grant the service principal permissions in Azure AD

Creating an External Volume for Google Cloud Storage

CREATE OR REPLACE EXTERNAL VOLUME iceberg_gcs_vol
  STORAGE_LOCATIONS = (
    (
      NAME = 'gcs-us-primary'
      STORAGE_BASE_URL = 'gcs://my-iceberg-bucket/warehouse/'
      STORAGE_PROVIDER = 'GCS'
    )
  );

-- After DESCRIBE, grant Storage Admin to the GCP service account

Cloud-by-Cloud Configuration Comparison

Creating Iceberg Tables Managed by the Snowflake Catalog

-- Create an Iceberg Table managed by the Snowflake catalog
-- Snowflake manages both metadata and data (read/write enabled)
CREATE OR REPLACE ICEBERG TABLE analytics.events.user_actions (
  event_id STRING,
  user_id INT,
  event_type STRING,
  event_timestamp TIMESTAMP_NTZ,
  properties OBJECT
)
  CATALOG = 'SNOWFLAKE'
  EXTERNAL_VOLUME = 'iceberg_s3_vol'
  BASE_LOCATION = 'analytics/user_actions/';

-- DML works just like a regular table
INSERT INTO analytics.events.user_actions
VALUES ('evt-001', 12345, 'page_view', '2026-03-27 10:00:00', PARSE_JSON('{"page": "/home"}'));

SELECT * FROM analytics.events.user_actions
WHERE event_timestamp >= '2026-03-27';

Creating Iceberg Tables Backed by an External Catalog

-- Integrate with AWS Glue Data Catalog
-- First create a Catalog Integration
CREATE OR REPLACE CATALOG INTEGRATION glue_catalog
  CATALOG_SOURCE = GLUE
  CATALOG_NAMESPACE = 'analytics_db'
  TABLE_FORMAT = ICEBERG
  GLUE_AWS_ROLE_ARN = 'arn:aws:iam::123456789012:role/snowflake-glue-role'
  GLUE_CATALOG_ID = '123456789012'
  GLUE_REGION = 'us-east-1'
  ENABLED = TRUE;

-- Iceberg Table managed by the external catalog (read-only)
CREATE OR REPLACE ICEBERG TABLE analytics.events.external_events
  EXTERNAL_VOLUME = 'iceberg_s3_vol'
  CATALOG = 'glue_catalog'
  CATALOG_TABLE_NAME = 'events';

Comparing Catalog Management Approaches

Managing External Volumes

-- List
SHOW EXTERNAL VOLUMES;

-- Inspect details (information required for IAM setup)
DESCRIBE EXTERNAL VOLUME iceberg_s3_vol;

-- Drop an External Volume (Iceberg Tables that reference it must be dropped first)
DROP EXTERNAL VOLUME iceberg_s3_vol;

Best Practices

• Separate External Volumes by team or domain: split External Volumes and buckets by use case for least-privilege IAM and cleaner cost attribution
• Use BASE_LOCATION to logically partition paths: when multiple tables share an External Volume, give each a distinct BASE_LOCATION
• Prefer the Snowflake catalog: use the Snowflake catalog when Snowflake reads and writes the data; reach for an external catalog only when you need multi-engine access
• Storage lifecycle rules: align bucket lifecycle rules with Iceberg snapshot expiration to archive or remove stale data files

Check Your Understanding

Frequently Asked Questions