Snowflake Event Tables - What Are Those?

In this article, we will be talking about observability, its importance and how Event tables can be used to improve observability and debugging.

In modern data engineering, observability is key to maintaining robust data platforms. Snowflake, a leading cloud data platform, introduced Event Tables to give developers and administrators deeper insight into application behaviour. These specialized tables provide automatic logging and tracing of activities within Snowflake stored procedures, UDFs, and Snowpark applications. This article explores the functionality of Snowflake Event Tables, their configuration, and how they enhance system observability and debugging efforts.

Features change from time to time with new features being added regularly, it is recommended that you review the documentation for the latest on what specific features are included with any of the Editions.

Open Table of Contents

What Are Snowflake Event Tables?
Setting Up Logging & Event Tables
Best Practices and Optimization Tips
Conclusion

What Are Snowflake Event Tables?

Snowflake Event Tables are special-purpose database tables designed to automatically collect and store telemetry data (logs and traces) from within your Snowflake environment.

Key characteristics include:

Predefined schema following the OpenTelemetry format.
Automatic log collection from stored procedures, UDFs, and Snowpark apps.
Centralized logging and tracing for better observability.
Queryable via SQL like standard tables.

By default, Snowflake provides SNOWFLAKE.TELEMETRY.EVENTS, a built-in event table. However, users can also create custom event tables for more granular control.

How Event Tables Work

Event Emission

Snowflake emits events from procedures and functions when logging and tracing APIs (like SYSTEM$LOG) are used. The telemetry emitted includes:

Log messages (INFO, WARN, ERROR, etc.)
Stack traces for errors
Custom span information for tracing

These are automatically inserted into the associated event table.

Predefined Schema

Event tables have a fixed schema that supports logs, spans, and resource metadata. Key columns include:

EVENT_TYPE: Type of event (LOG, SPAN, etc.)
BODY: Message body or stack trace
TIMESTAMP: UTC timestamp of the event
SEVERITY: Log severity (e.g., INFO, WARN)
RESOURCE_ATTRIBUTES: Metadata like user, session, function name

This structure aligns with OpenTelemetry for easy integration with external tools.

Scoping and Association

Event tables can be associated with:

The account level using:

ALTER ACCOUNT SET EVENT_TABLE = <DATABASE>.<SCHEMA>.<EVENT_TABLE>

A specific database, schema, or procedure

This flexibility allows selective logging depending on your observability needs.

Setting Up Logging & Event Tables

Enabling Logging with Default Table

To use the default Snowflake event table:

Set your telemetry level:

ALTER ACCOUNT SET LOG_LEVEL = 'INFO';

Use logging functions in your procedure:

CALL SYSTEM$LOG('INFO', 'Starting ETL step');

Query log entries:

SELECT * FROM SNOWFLAKE.TELEMETRY.EVENTS
WHERE RESOURCE_ATTRIBUTES:procedure_name = 'my_proc';

Creating a Custom Event Table

CREATE EVENT TABLE my_database.my_schema.custom_events;

Associating the Table

ALTER DATABASE my_database SET EVENT_TABLE = my_database.my_schema.custom_events;

Or at the account level:

ALTER ACCOUNT SET EVENT_TABLE = my_database.my_schema.custom_events;

Access Control

Use these predefined roles to manage permissions:

SNOWFLAKE.EVENTS_ADMIN – full control

SNOWFLAKE.EVENTS_VIEWER – read-only access to EVENTS_VIEW

GRANT APPLICATION ROLE SNOWFLAKE.EVENTS_VIEWER TO ROLE analyst_role;

Best Practices and Optimization Tips

Use custom event tables per environment (e.g., dev, prod) for isolation.
Filter logs by severity and context to avoid noisy data.
Leverage EVENTS_VIEW to apply row access policies.

Conclusion

As you can see, the options are limitless, using event tables to log and provide observability in stored procedures, debugging errors, providing users with audit trails and performance monitoring.

Snowflake Event Tables provide a powerful, native approach to logging and observability within your Snowflake workloads. Whether you’re debugging stored procedures, auditing application activity, or building real-time monitoring dashboards, event tables give you structured, queryable access to what’s happening under the hood. With careful setup and role-based controls, you can maintain system transparency while controlling costs and access.