Snowflake Integration
Last updated
Last updated
Snowflake Enterprise Edition required
In this integration, Immuta manages access to Snowflake tables by administering Snowflake row access policies and column masking policies on those tables, allowing users to query tables directly in Snowflake while dynamic policies are enforced.
Like with all Immuta integrations, Immuta can inject its ABAC model into policy building and administration to remove policy management burden and significantly reduce role explosion.
When an administrator configures the Snowflake integration with Immuta, Immuta creates an IMMUTA database and schemas (immuta_procedures, immuta_policies, and immuta_functions) within Snowflake to contain policy definitions and user entitlements. Immuta then creates a system role and gives that system account the privileges required to orchestrate policies in Snowflake and maintain state between Snowflake and Immuta. See the Snowflake privileges section for a list of privileges, the user they must be granted to, and an explanation of why they must be granted.
An Immuta application administrator configures the Snowflake integration and registers Snowflake warehouse and databases with Immuta.
Immuta creates a database inside the configured Snowflake warehouse that contains Immuta policy definitions and user entitlements.
A data owner registers Snowflake tables in Immuta as data sources.
If Snowflake tag ingestion was enabled during the configuration, Immuta uses the host provided in the configuration and ingests internal tags on Snowflake tables registered as Immuta data sources.
A data owner, data governor, or administrator creates or changes a policy or a user's attributes change in Immuta.
The Immuta web service calls a stored procedure that modifies the user entitlements or policies.
Immuta manages and applies Snowflake governance column and row access policies to Snowflake tables that are registered as Immuta data sources.
If Snowflake table grants is not enabled, Snowflake object owner or user with the global MANAGE GRANTS privilege grants SELECT privilege on relevant Snowflake tables to users. Note: Although they are GRANTed access, if they are not subscribed to the table via Immuta-authored policies, they will not see data.
A Snowflake user who is subscribed to the data source in Immuta queries the corresponding table directly in Snowflake and sees policy-enforced data.
When Immuta users create policies, they are then pushed into the Immuta database within Snowflake; there, the Immuta system account orchestrates Snowflake row access policies and column masking policies directly onto Snowflake tables. Changes in Immuta policies, user attributes, or data sources trigger webhooks that keep the Snowflake policies up-to-date.
For a user to query Immuta-protected data, they must meet two qualifications:
They must be subscribed to the Immuta data source.
They must be granted SELECT access on the table by the Snowflake object owner or automatically via the Snowflake table grants feature.
After a user has met these qualifications they can query Snowflake tables directly.
See the integration support matrix on the Data policy types reference guide for a list of supported data policy types in Snowflake.
When a user applies a masking policy to a Snowflake data source, Immuta truncates masked values to align with Snowflake column length (VARCHAR(X) types) and precision (NUMBER (X,Y) types) requirements.
Consider these columns in a data source that have the following masking policies applied:
Column A (VARCHAR(6)): Mask using hashing for everyone
Column B (VARCHAR(5)): Mask using a constant REDACTED for everyone
Column C (VARCHAR(6)): Mask by making null for everyone
Column D (NUMBER(3, 0)): Mask by rounding to the nearest 10 for everyone
Querying this data source in Snowflake would return the following values:
5w4502
REDAC
990
6e3611
REDAC
750
9s7934
REDAC
380
Hashing collisions
Hashing collisions are more likely to occur across or within Snowflake columns restricted to short lengths, since Immuta truncates the hashed value to the limit of the column. (Hashed values truncated to 5 characters have a higher risk of collision than hashed values truncated to 20 characters.) Therefore, avoid applying hashing policies to Snowflake columns with such restrictions.
For more details about Snowflake column length and precision requirements, see the Snowflake behavior change release documentation.
When a policy is applied to a column, Immuta uses Snowflake memoizable functions to cache the result of the called function. Then, when a user queries a column that has that policy applied to it, Immuta uses that cached result to dramatically improve query performance.
The privilege grants the Snowflake integration requires align to the least privilege security principle. The table below describes each privilege required in Snowflake for the or the user. The references to IMMUTA_DB , IMMUTA_WH, and IMMUTA_IMPERSONATOR_ROLE in the table can be replaced with what you chose for the name of your Immuta database, warehouse, and impersonation role when setting up the integration, respectively.
CREATE DATABASE ON ACCOUNT WITH GRANT OPTION
Setup user
All
The setup script this user runs creates an Immuta database in the customer Snowflake account where all Immuta managed objects (UDFs, masking policies, row access policies, and user entitlements) will be written and stored.
CREATE ROLE ON ACCOUNT WITH GRANT OPTION
Setup user
All
The setup script this user runs creates a ROLE for Immuta that will be used to manage the integration once it has been initialized.
CREATE USER ON ACCOUNT WITH GRANT OPTION
Setup user
All
The setup script this user runs creates the IMMUTA_SYSTEM_ACCOUNT user that Immuta will use to manage the integration.
MANAGE GRANTS ON ACCOUNT
Setup user
All
The user configuring the integration must be able to GRANT global privileges and access to objects within the Snowflake account. All privileges that are documented here are granted to the IMMUTA_SYSTEM_ACCOUNT user by this setup user.
OWNERSHIP ON ROLE IMMUTA_IMPERSONATOR_ROLE
IMMUTA_SYSTEM_ACCOUNT user
Impersonation
If impersonation is enabled, Immuta must be able to manage the Snowflake roles used for impersonation, which is created when the setup script runs, in order to manage the impersonation feature.
ALL PRIVILEGES ON DATABASE IMMUTA_DB
ALL PRIVILEGES ON ALL SCHEMAS IN DATABASE IMMUTA_DB
USAGE ON FUTURE PROCEDURES IN SCHEMA IMMUTA_DB.IMMUTA_PROCEDURES
IMMUTA_SYSTEM_ACCOUNT user
All
The setup script grants the Immuta system account user these privileges because Immuta must have full ownership of the Immuta database where Immuta objects are managed.
USAGE ON WAREHOUSE IMMUTA_WH
IMMUTA_SYSTEM_ACCOUNT user
All
To make changes to state in the Immuta database, Immuta requires access to compute (a Snowflake warehouse). Some state changes are DDL operations, and others are DML and require compute.
IMPORTED PRIVILEGES ON DATABASE SNOWFLAKE
IMMUTA_SYSTEM_ACCOUNT user
Audit
APPLY TAG ON ACCOUNT
IMMUTA_SYSTEM_ACCOUNT user
Tag ingestion
APPLY MASKING POLICY ON ACCOUNT
APPLY ROW ACCESS POLICY ON ACCOUNT
IMMUTA_SYSTEM_ACCOUNT user
Snowflake integration with governance features enabled
MANAGE GRANTS ON ACCOUNT
IMMUTA_SYSTEM_ACCOUNT user
Table grants
Immuta must be able to MANAGE GRANTS on objects throughout the customer's Snowflake account.
CREATE ROLE ON ACCOUNT
IMMUTA_SYSTEM_ACCOUNT user
Table grants
When using the table grants feature, Immuta must be able to create roles as targets for Immuta subscription policy permissions in the customer’s Snowflake account.
USAGE ON DATABASE IMMUTA_DB
USAGE ON SCHEMA IMMUTA_DB.IMMUTA_PROCEDURES
USAGE ON SCHEMA IMMUTA_DB.IMMUTA_FUNCTIONS
USAGE ON FUTURE FUNCTIONS IN SCHEMA IMMUTA_DB.IMMUTA_FUNCTIONS
USAGE ON SCHEMA IMMUTA_DB.IMMUTA_SYSTEM
SELECT ON IMMUTA_DB.IMMUTA_SYSTEM.USER_PROFILE
PUBLIC role
All
Immuta has stored procedures and functions that are used for policy enforcement and do not expose or contain any sensitive information. These objects must be accessible by all users to facilitate the use and creation of policies or views to enforce Immuta policies in Snowflake.
SELECT ON IMMUTA_DB.IMMUTA_SYSTEM.ALLOW_LIST
PUBLIC role
All
Immuta retains a list of excepted roles and users when using the Snowflake integration. The roles and users in this list will be exempt from policies applied to tables in Snowflake to give customers flexibility in case there are entities that should not be bound to Immuta policies in Snowflake (for example, a system or application role or user).
Register Snowflake data sources using a dedicated Snowflake role. Avoid using individual user accounts for data source onboarding. Instead, create a service account (Snowflake user account TYPE=SERVICE) with SELECT access for onboarding data sources. No policies will apply to that role, ensuring that your integration works with the following use cases:
Snowflake project workspaces: Snowflake workspaces generate static views with the credentials used to register the table as an Immuta data source. Those tables must be registered in Immuta by an excepted role so that policies applied to the backing tables are not applied to the project workspace views.
Using views and tables within Immuta: Because this integration uses Snowflake governance policies, users can register tables and views as Immuta data sources. However, if you want to register views and apply different policies to them than their backing tables, the owner of the view must be an excepted role; otherwise, the backing table’s policies will be applied to that view.
Private preview: This feature is only available to select accounts. Reach out to your Immuta representative to enable this feature.
Bulk data source creation is the more efficient process when loading more than 5000 data sources from Snowflake and allows for data sources to be registered in Immuta before running sensitive data discovery or applying policies.
To use this feature, see the Bulk create Snowflake data sources guide.
Based on performance tests that create 100,000 data sources, the following minimum resource allocations need to be applied to the appropriate pods in your Kubernetes environment for successful bulk data source creation.
Memory
4Gi
16Gi
CPU
2
4
Storage
8Gi
24Gi
Performance gains are limited when enabling sensitive data discovery at the time of data source creation.
External catalog integrations are not recognized during bulk data source creation. Users must manually trigger a catalog sync for tags to appear on the data source through the data source's health check.
Excepted roles and users are assigned when the integration is installed, and no policies will apply to these users' queries, despite any Immuta policies enforced on the tables they are querying. Credentials used to register a data source in Immuta will be automatically added to this excepted list for that Snowflake table. Consequently, roles and users added to this list and used to register data sources in Immuta should be limited to service accounts.
Immuta excludes the listed roles and users from policies by wrapping all policies in a CASE statement that will check if a user is acting under one of the listed usernames or roles. If a user is, then the policy will not be acted on the queried table. If the user is not, then the policy will be executed like normal. Immuta does not distinguish between role and username, so if you have a role and user with the exact same name, both the user and any user acting under that role will have full access to the data sources and no policies will be enforced for them.
The Snowflake integration supports the following authentication methods to configure the integration and create data sources:
Username and password: Users can authenticate with their Snowflake username and password.
Key pair: Users can authenticate with a Snowflake key pair authentication.
Snowflake External OAuth: Users can authenticate with Snowflake External OAuth.
Immuta's OAuth authentication method uses the Client Credentials Flow to integrate with Snowflake External OAuth. When a user configures the Snowflake integration or connects a Snowflake data source, Immuta uses the token credentials (obtained using a certificate or passing a client secret) to craft an authenticated access token to connect with Snowflake. This allows organizations that already use Snowflake External OAuth to use that secure authentication with Immuta.
An Immuta application administrator configures the Snowflake integration or creates a data source.
Immuta creates a custom token and sends it to the authorization server.
The authorization server confirms the information sent from Immuta and issues an access token to Immuta.
Immuta sends the access token it received from the authorization server to Snowflake.
Snowflake authenticates the token and grants access to the requested resources from Immuta.
The integration is connected and users can query data.
The Immuta Snowflake integration supports Snowflake external tables. However, you cannot add a masking policy to an external table column while creating the external table in Snowflake because masking policies cannot be attached to virtual columns.
The Snowflake integration supports the Immuta features outlined below. Click the links provided for more details.
Immuta project workspaces: Users can have additional write access in their integration using project workspaces.
Tag ingestion: Immuta automatically ingests Snowflake object tags from your Snowflake instance and adds them to the appropriate data sources.
User impersonation: Native impersonation allows users to natively query data as another Immuta user. To enable native user impersonation, see the Integration user impersonation page.
Native query audit: Immuta audits queries run natively in Snowflake against Snowflake data registered as Immuta data sources.
Snowflake low row access policy mode: The Snowflake low row access policy mode improves query performance in Immuta's Snowflake integration by decreasing the number of Snowflake row access policies Immuta creates.
Snowflake table grants: This feature allows Immuta to manage privileges on your Snowflake tables and views according to the subscription policies on the corresponding Immuta data sources.
Immuta system account required Snowflake privileges
CREATE [OR REPLACE] PROCEDURE
DROP ROLE
REVOKE ROLE
Users can have additional write access in their integration using project workspaces. For more details, see the Snowflake project workspaces page.
To use project workspaces with the Snowflake integration, the default role of the account used to create data sources in the project must be added to the "Excepted Roles/Users List." If the role is not added, you will not be able to query the equalized view using the project role in Snowflake.
You can enable Snowflake tag ingestion so that Immuta will ingest Snowflake object tags from your Snowflake instance into Immuta and add them to the appropriate data sources.
The Snowflake tags' key and value pairs will be reflected in Immuta as two levels: the key will be the top level and the value the second. As Snowflake tags are hierarchical, Snowflake tags applied to a database will also be applied to all of the schemas in that database, all of the tables within those schemas, and all of the columns within those tables. For example: If a database is tagged PII, all of the tables and columns in that database will also be tagged PII.
To enable Snowflake tag ingestion, see the Configure a Snowflake integration page.
Snowflake has some natural data latency. If you manually refresh the governance page to see all tags created globally, users can experience a delay of up to two hours. However, if you run schema detection or a health check to find where those tags are applied, the delay will not occur because Immuta will only refresh tags for those specific tables.
Immuta system account required Snowflake privilege
IMPORTED PRIVILEGES ON DATABASE snowflake
Once this feature has been enabled with the Snowflake integration, Immuta will query Snowflake to retrieve user query histories. These histories provide audit records for queries against Snowflake data sources that are queried natively in Snowflake.
This process will happen automatically every hour by default but can be configured to a different frequency when configuring or editing the integration. Additionally, audit ingestion can be manually requested at any time from the Immuta audit page. When manually requested, it will only search for new queries that were created since the last native query that had been audited. The job is run in the background, so the new queries will not be immediately available.
For details about prompting these logs and the contents of these audit logs, see the Snowflake query audit logs page.
A user can configure multiple integrations of Snowflake to a single Immuta tenant and use them dynamically or with workspaces.
There can only be one integration connection with Immuta per host.
The host of the data source must match the host of the integration for the view to be created.
Projects can only be configured to use one Snowflake host.
If there are errors in generating or applying policies natively in Snowflake, the data source will be locked and only users on the excepted roles/users list and the credentials used to create the data source will be able to access the data.
Once a Snowflake integration is disabled in Immuta, the user must remove the access that was granted in Snowflake. If that access is not revoked, users will be able to access the raw table in Snowflake.
Migration must be done using the credentials and credential method (automatic or bootstrap) used to configure the integration.
When configuring one Snowflake instance with multiple Immuta tenants, the user or system account that enables the integration on the app settings page must be unique for each Immuta tenant.
A Snowflake table can only have one set of policies enforced at a given time, so creating multiple data sources pointing to the same table is not supported. If this is a use case you need to support, create views in Snowflake and expose those instead.
You cannot add a masking policy to an external table column while creating the external table because a masking policy cannot be attached to a virtual column.
If you create an Immuta data source from a Snowflake view created using a select * from query, Immuta column detection will not work as expected because Snowflake views are not automatically updated based on backing table changes. To remedy this, you can create views that have the specific columns you want or you can CREATE AND REPLACE the view in Snowflake whenever the backing table is updated and manually run the column detection job on the data source page.
If a user is created in Snowflake after that user is already registered in Immuta, Immuta does not grant usage on the per-user role automatically - meaning Immuta does not govern this user's access without manual intervention. If a Snowflake user is created after that user is registered in Immuta, the user account must be disabled and re-enabled to trigger a sync of Immuta policies to govern that user. Whenever possible, Snowflake users should be created before registering those users in Immuta.
Snowflake tables from imported databases are not supported. Instead, create a view of the table and register that view as a data source.
The Immuta Snowflake integration uses Snowflake governance features to let users query data natively in Snowflake. This means that Immuta also inherits some Snowflake limitations using correlated subqueries with row access policies and column-level security. These limitations appear when writing custom WHERE policies, but do not remove the utility of row-level policies.
All column names must be fully qualified: Any column names that are unqualified (i.e., just the column name) will default to a column of the data source the policy is being applied to (if one matches the name).
The Immuta system account must have SELECT privileges on all tables/views referenced in a subquery: The Immuta system role name is specified by the user, and the role is created when the Snowflake instance is integrated.
Any subqueries that error in Snowflake will also error in Immuta.
Including one or more subqueries in the Immuta policy condition may cause errors in Snowflake. If an error occurs, it may happen during policy creation or at query-time. To avoid these errors, limit the number of subqueries, limit the number of JOIN operations, and simplify WHERE clause conditions.
For more information on the Snowflake subquery limitations see
To ingest audit information from Snowflake, Immuta must have access to the SNOWFLAKE.ACCOUNT_USAGE.ACCESS_HISTORY view. See the for details.
To ingest table, view, and column tag information from Snowflake, Immuta must have this permission. Immuta reads from the TAG_REFERENCES .
Immuta must be able to apply policies to objects throughout the customer’s Snowflake account and query for existing policies on objects using the POLICY_REFERENCES .
