Manage Data Shared With the RAI Native App

RelationalAI (RAI) Python users build models on top of data stored in Snowflake tables and views. Before models can be queried, the data must be shared with the RAI Native App using a data stream. These data streams are maintained by the RAI Native App’s CDC Service.

Note

Support for data streams is disabled by default when you install the RAI Native App. You must enable the CDC service before data streams can be created.

The CDC Service

Data streams use change data capture (CDC) to stream updates from Snowflake tables and views to the RAI Native App. The CDC service processes the change tracking data consumed by data streams to keep RAI models synchronized with their source data.

Note

The CDC service requires that your RAI Native App be granted the EXECUTE TASK and EXECUTE MANAGED TASK privileges. This is typically done when you install the app.

You can manage the CDC service using SQL, Python, or the RAI CLI.

Enable CDC

Requires the cdc_admin application role.

SQL

To enable the CDC service, or to resume the service after suspending it, use the app.resume_cdc() procedure:

-- Enable the CDC service.
CALL relationalai.app.resume_cdc();
/*+--------------------------------------------------------------------+
  | CDC functionality on the RelationalAI application has been resumed |
  +--------------------------------------------------------------------+ */

Python

To enable the CDC service, or to resume the service after suspending it, create a Provider instance and use its .sql() method to execute the app.resume_cdc() SQL procedure:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Enable the CDC service.
app.sql("CALL relationalai.app.resume_cdc()")

CLI

To enable the CDC service, or to resume the service after suspending it, use the imports:setup command’s --resume flag:

# Enable the CDC service.
rai imports:setup --resume

Note

While CDC is enabled, a dedicated engine responsible for processing changes tracked by data streams is provisioned whenever changes are detected. See CDC Engine for details.

Disable CDC

Requires the cdc_admin application role.

SQL

To disable the CDC service, use the app.suspend_cdc() procedure:

-- Disable the CDC service.
CALL relationalai.app.suspend_cdc();
/*+-----------------------------------------------------------------------------------------------------------------+
  | CDC functionality on the RelationalAI application has been suspended and its associated engine has been dropped |
  +-----------------------------------------------------------------------------------------------------------------+ */

Python

To disable the CDC service, create a Provider instance and use its .sql() method to execute the app.suspend_cdc() procedure:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Disable the CDC service.
app.sql("CALL relationalai.app.suspend_cdc()")

CLI

To disable the CDC service, use the imports:setup command’s --suspend flag:

# Disable the CDC service.
rai imports:setup --suspend

Disabling CDC suspends the CDC engine. Change tracking data is still consumed by data streams, but is not processed until the service is resumed. Data streams cannot be created while CDC is disabled.

Note

When you disable the CDC service, RAI Python models retain access to a snapshot of the source data. Model queries are evaluated against the snapshot.

View CDC Service Status

Requires the app_user application role.

SQL

To view the status of the CDC service, use the app.cdc_status() procedure:

-- Get the CDC service status.
CALL relationalai.app.cdc_status();
/*+--------------+--------------------+-------------------+------------------+-----------------+--------------------------------------------------------------------------------------------------------------------------|
  | CDC_ENABLED  | CDC_ENGINE_NAME    | CDC_ENGINE_STATUS | CDC_ENGINE_SIZE  | CDC_TASK_STATUS | CDC_TASK_INFO                                                                                                            |
  |--------------+--------------------+-------------------+------------------+-----------------+--------------------------------------------------------------------------------------------------------------------------|
  | TRUE         | CDC_MANAGED_ENGINE | READY             | HIGHMEM_X64_S    | started         | {"createdOn": "2024-10-15 21:58:11.291 -0700", "lastSuspendedOn": null, "lastSuspendedReason": null, "state": "started"} |
  +--------------+--------------------+-------------------+------------------+-----------------+--------------------------------------------------------------------------------------------------------------------------| */

Refer to the reference docs for more details on the output of the cdc_status() procedure.

Python

To view the status of the CDC service, create a Provider instance and use its .sql() method to execute the app.cdc_status() procedure:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Get the CDC service status.
cdc_status = app.sql("CALL relationalai.app.cdc_status()")
print(cdc_status)
# [Row(CDC_ENABLED=True, CDC_ENGINE_NAME='CDC_MANAGED_ENGINE', CDC_ENGINE_STATUS='READY', CDC_ENGINE_SIZE='M', CDC_TASK_STATUS='started', CDC_TASK_INFO='{\n  "createdOn": "2024-10-24 22:53:15.200 -0700",\n  "lastSuspendedOn": null,\n  "lastSuspendedReason": null,\n  "state": "started"\n}')]

CLI

To view the status of the CDC service, use the imports:setup command:

$ rai imports:setup

---------------------------------------------------

▰▰▰▰ Imports setup fetched

To suspend imports, use 'rai imports:setup --suspend'


  Field                 Value
 ───────────────────────────────────────────
  engine                CDC_MANAGED_ENGINE
  engine_size           M
  engine_status         READY
  status                STARTED
  enabled               True
  createdOn             2024-10-24 22:53:15
  lastSuspendedOn       N/A
  lastSuspendedReason   N/A


---------------------------------------------------

Configure CDC Engine Size

Requires the cdc_admin application role.

When the RAI Native App is installed, the CDC engine is configured with the HIGHMEM_X64_S size. If you encounter performance issues with the CDC engine or need to process large data streams, you can change the CDC engine size to a larger instance family. See Engine Sizes for more information on available engine sizes.

SQL

To change the size of the CDC engine, use the app.alter_cdc_engine_size() procedure:

-- Change the size of the CDC engine to HIGHMEM_X64_M.
CALL relationalai.app.alter_cdc_engine_size('HIGHMEM_X64_M');
/*+--------------------------------------+
  | CDC engine size set to HIGHMEM_X64_M |
  +--------------------------------------+ */

Python

To change the size of the CDC engine, create a Provider instance and use its .sql() method to execute the app.alter_cdc_engine_size() procedure:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Change the size of the CDC engine to HIGHMEM_X64_M.
app.sql("CALL relationalai.app.alter_cdc_engine_size('HIGHMEM_X64_M')")

CLI

To change the size of the CDC engine, pass the new engine size to the imports:setup command’s --engine_size flag:

# Change the size of the CDC engine to HIGHMEM_X64_M.
rai imports:setup --engine_size HIGHMEM_X64_M

If a batch of data stream changes is currently being processed, it is completed using the previously configured engine. A new engine with the new size is created whenever the next batch of changes is processed, at which point the old engine is deleted.

Data Streams

Data streams track changes to tables and views in your Snowflake account, ensuring that queries from RAI Python models use the latest data. Changes are batched and processed by the CDC Service every minute.

The RAI Python API automatically creates data streams. Python users can populate an RAI Type from rows in a source table or view they have access to. When a query is executed for the first time, data streams are created for the source tables or views if they do not already exist.

Although data streams can be created manually using SQL or the RAI CLI, the recommended approach is to use the Python API, which handles stream creation automatically.

Note

You must enable the CDC service before data streams can be created.

Supported Source Types

Data streams can be created on Snowflake tables and views, with some limitations:

• The user creating a data stream must have SELECT privileges on the source table or view.
• The source table or view must have change tracking enabled.
• Data streams cannot be created on temporary or transient tables.
• Data streams are not supported on external tables or views.

Refer to the Snowflake documentation for more details on stream requirements and limitations.

Supported Column Types

Data streams support the following Snowflake column types:

• VARCHAR
• BINARY
• NUMBER
• FLOAT
• BOOLEAN
• DATE
• TIME
• TIMESTAMP_NTZ
• TIMESTAMP_LTZ
• TIMESTAMP_TZ

Tables or views with unsupported column types cannot be streamed into the RAI Native App.

Note

If you need to stream data from a source object with unsupported column types, consider creating a view without those columns or that casts the unsupported columns to a supported type.

Change Tracking

Data streams use Snowflake’s native change tracking feature to capture changes to tables and views. You must enable change tracking on a source table or view before you can create a data stream.

Here’s how it works:

• Change tracking captures information about all Data Manipulation Language (DML) statements committed to a table or view, including INSERT, UPDATE, and DELETE operations.
• Data streams read these changes and temporarily store them in a internal stage inside the RAI Native App’s Snowflake database. This data is staged even if the CDC Service is disabled.
• When the CDC Service is enabled, the staged data is processed in batches once every minute and made available to RAI models.

Staleness

When a stream is suspended, it stops consuming change tracking data from its source table or view, and that data is no longer stored in the app’s Snowflake database. Suspending a stream can help reduce storage costs, but Snowflake retains change tracking data only for a limited time, after which it becomes inaccessible.

If a stream remains suspended beyond its data retention period, it becomes stale. A stale stream may fail to catch up with recent changes when resumed, leading to data inconsistencies.

Tip

Resume suspended data streams periodically to avoid staleness. Note that the length of a stream’s data retention period is configurable. See the Snowflake documentation for details.

If you encounter data inconsistencies due to staleness, you can delete the stream and recreate it to start fresh.

Quarantined Streams

If the CDC Service encounters too many failures while attempting to process a data stream, the stream is given a QUARANTINED status. Change tracking data for quarantined streams is consumed, but not processed. Resolve the errors causing the stream to be quarantined, then resume the data stream to process the pending changes. See View Quarantined Streams for details on viewing errors for quarantined streams.

Security and Access Control

The RAI Native App uses Snowflake’s security features to manage access to data streams and the underlying source tables and views.

Role-Based Access Control (RBAC)

RAI supports Snowflake’s Role-Based Access Control (RBAC) model, which manages access to data and resources through roles and privileges. In Snowflake, privileges like SELECT or USAGE are granted to roles, and roles are assigned to users. This allows organizations to control who can access specific tables, views, and other objects, and what actions they can perform.

RAI enforces Snowflake RBAC as follows:

• All app operations use the privileges of the active roles in your Snowflake session, as set in your configuration.
• When you create or use a data stream, Snowflake checks that your role has the required privileges on the object and its parent database and schema.
• If privileges are missing, access is denied.

Fine-Grained Access Control

Users with the Enterprise Edition of Snowflake (including Business Critical and Virtual Private Snowflake) can use row access policies for controlling which rows are visible and masking policies for hiding sensitive column data.

When the RAI Native App executes SQL commands, all row and masking policies are evaluated against a role named RELATIONALAI—not the role defined in the user’s configuration .

This means that:

• RAI currently does not support user-based policies at the row and column level.
• You can create policies that apply to the RELATIONALAI role, but these policies impact all users of the app.

Practical Example

Consider the following masking policy that masks sensitive employee passwords when the user does not have the FULL_ACCESS role:

CREATE MASKING POLICY employee_pwd_mask AS (val string) RETURNS string ->
  CASE
    WHEN CURRENT_ROLE() IN ('FULL_ACCESS') THEN val
    ELSE '******'
  END;

The app executes all queries using the RELATIONALAI role, so this masking policy will apply to all users of the app. All users will see the masked value, even if the user has the FULL_ACCESS role.

Now suppose you add RELATIONALAI to the list of roles that can see the original value:

CREATE MASKING POLICY employee_pwd_mask AS (val string) RETURNS string ->
  CASE
    WHEN CURRENT_ROLE() IN ('FULL_ACCESS', 'RELATIONALAI') THEN val
    ELSE '******'
  END;

In this case, the app will see the original value of the password and so will any app user with SELECT privileges on the relevant table, even if they do not have the FULL_ACCESS role.

Best Practices

For data access policies to work effectively with the RAI Native App, consider the following best practices:

• Start with least privilege: Deny access by default, then allow only what the app needs.
• Write policies that check the app’s role, not individual users.
• If you use mapping tables for access control, map to the app’s role—not individual users.

Data Stream Management

You can manage data streams using SQL, Python, or the RAI CLI.

Enable Change Tracking on a Table or View

Requires ownership privileges on the table or view.

SQL

To enable change tracking on a table or view, use the ALTER TABLE statement:

-- Enable change tracking on a table.
ALTER TABLE MyTable SET CHANGE_TRACKING = TRUE;

-- Enable change tracking on a view.
ALTER VIEW MyView SET CHANGE_TRACKING = TRUE;

-- Verify that change tracking is enabled. If change tracking is enabled, the
-- CHANGE_TRACKING column will be set to ON.
SHOW TABLES LIKE 'MyTable';

Python

To enable change tracking on a table or view, create a Provider instance and use its .sql() method to execute the ALTER TABLE statement:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Enable change tracking on a table.
app.sql("ALTER TABLE MyTable SET CHANGE_TRACKING = TRUE")

# Enable change tracking on a view.
app.sql("ALTER VIEW MyView SET CHANGE_TRACKING = TRUE")

# Verify that change tracking is enabled. If change tracking is enabled, the
# CHANGE_TRACKING column will be set to ON.
print(app.sql("SHOW TABLES LIKE 'MyTable'"))

See Change Tracking for details on what operations are tracked.

Create a Data Stream

Requires the cdc_admin application role.

SQL

To create a data stream, use the api.create_data_stream() procedure:

-- Replace the placeholders with your database, schema, and table/view names.
SET obj_name = '<db>.<schema>.<table_or_view>';
SET obj_type = 'TABLE';  -- Set to 'VIEW' if needed.
SET obj_ref = relationalai.api.object_reference($obj_type, $obj_name);

-- Enable change tracking on the table or view
ALTER TABLE IDENTIFIER($obj_name) SET CHANGE_TRACKING = TRUE;

-- Create a data stream named 'my_stream' for the object reference
CALL relationalai.api.create_data_stream($obj_ref, 'my_stream', TRUE);
/*+----------------------------------+
  | Datastream created successfully. |
  +----------------------------------+ */

Python

To create a data stream, create a Provider instance and use its .create_streams() method:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

table_or_view = "<db>.<schema>.<table_or_view>"

# Enable change tracking on the table or view.
app.sql(f"ALTER TABLE IDENTIFIER('{table_or_view}') SET CHANGE_TRACKING = TRUE;")

# Create a data stream named 'my_stream' for the table or view.
app.create_streams([table_or_view], 'MyModel')

CLI

To create a data stream, pass the fully-qualified table or view name and the model name to the imports:stream command’s --source and --model flags:

rai imports:stream --source <db>.<schema>.<table_or_view> --model MyModel

Note that change tracking must be enabled on the source table or view before creating a data stream. Not all column types are supported by data streams. See Supported Column Types for details.

Caution

You may use up to 1000 tables and 1000 views as source objects for data streams, for maximum total of 2000 source objects.

Suspend or Resume a Data Stream

Requires the cdc_admin application role.

SQL

To suspend a data stream, use the api.suspend_data_stream() procedure:

-- Suspend a data stream. Replace the placeholders with your database, schema,
-- and table or view name.
CALL relationalai.api.suspend_data_stream('<db>.<schema>.<table_or_view>');
/*+-----------------------+
  | Data stream suspended |
  +-----------------------+ */

While a data stream is suspended, the RAI Native App no longer consumes change tracking data for the stream’s source table or view. Suspended streams should be resumed at regular intervals to avoid becoming stale.

Note

When you suspend a data stream, RAI Python models retain access to a snapshot of the source data. Queries from models that consume data from a suspended stream are evaluated against the snapshot.

To resume a suspended data stream, use the api.resume_data_stream() procedure:

-- Resume a data stream. Replace the placeholders with your database, schema,
-- and table or view name.
CALL relationalai.api.resume_data_stream('<db>.<schema>.<table_or_view>');
/*+---------------------+
  | Data stream resumed |
  +---------------------+ */

Python

To suspend or resume a data stream, create a Provider instance and use its .sql() method to execute the api.suspend_data_stream() SQL procedure:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Suspend a data stream. Replace the placeholders with your database, schema,
# and table or view name.
app.sql("CALL relationalai.api.suspend_data_stream('<db>.<schema>.<table_or_view>')")

While a data stream is suspended, the RAI Native App no longer consumes change tracking data for the stream’s source table or view. Suspended streams should be resumed at regular intervals to avoid becoming stale.

Note

When you suspend a data stream, RAI Python models retain access to a snapshot of the source data. Queries from models that consume data from a suspended stream are evaluated against the snapshot.

To resume a suspended data stream, use the resume_data_stream() procedure:

# Resume a data stream. Replace the placeholders with your database, schema,
# and table or view name.
app.sql("CALL relationalai.api.resume_data_stream('<db>.<schema>.<table_or_view>')")

Delete a Data Stream

Requires the cdc_admin application role.

SQL

To delete a data stream, use the api.delete_data_stream() procedure:

-- Delete a data stream. Replace the placeholders with your database, schema, and table or view name.
CALL relationalai.api.delete_data_stream('<db>.<schema>.<table_or_view>');
/*+-----------------------------------+
  | Data stream deleted successfully. |
  +-----------------------------------+ */

Python

To delete a data stream, create a Provider instance and use its .delete_stream() method:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Delete a data stream. Replace the placeholders with your database, schema,
# and table or view name.
app.delete_stream('<db>.<schema>.<table_or_view>')

CLI

To delete a data stream, pass the fully-qualified table or view name and the model name to the imports:delete command’s --object and --model flags:

rai imports:delete --object <db>.<schema>.<table_or_view> --model MyModel

Note

When you delete a data stream, RAI Python models retain access to a snapshot of the source data. Queries from models that consume data from a deleted stream are evaluated against the snapshot.

List Data Streams

Requires the cdc_admin application role.

SQL

To view a list all data streams, query the api.data_streams view:

SELECT * FROM relationalai.api.data_streams;
/*+-------------------------------------+-------------------------+--------------------------+----------+-------------------+--------------------------------------+------------------------- -+--------------+--------------------------+
  | ID                                  | CREATED_AT              | CREATED_BY               | STATUS   | REFERENCE_NAME    | REFERENCE_ALIAS                      | FQ_OBJECT_NAME            | RAI_DATABASE | RAI_RELATION             |
  |-------------------------------------+-------------------------+--------------------------+----------+-------------------+--------------------------------------+---------------------------+--------------+--------------------------|
  | ds_a1b2c3d4_e5f6_7a89_b123_d456e789 | 2024-10-23 12:23:45.250 | john.doe@company.com     | CREATED  | DATA_STREAM_VIEW  | 1234abcd-5678-90ef-ab12-3456cdef7890 | example_db.sales.view1    | SalesModel   | example_db.sales.view1   |
  | ds_8e7f6d5c_4a3b_2c1d_0e9f_7b6a8d9f | 2024-10-22 15:37:29.580 | maria.garcia@company.com | CREATED  | DATA_STREAM_TABLE | bcd123ef-4567-890a-bcde-abcdef678901 | example_db.hr.employees   | HRModel      | example_db.hr.employees  |
  | ds_9a8b7c6d_5e4f_3d2a_1b0e_f7g6h5i3 | 2024-10-21 17:44:10.300 | mark.jones@company.com   | DELETING | DATA_STREAM_VIEW  | 7890abcd-1234-5678-90ef-bcde4567890f | example_db.finance.budget | FinanceModel | example_db.finance.budget|
  +-------------------------------------+-------------------------+--------------------------+----------+-------------------+--------------------------------------+---------------------------+--------------+--------------------------+ */

Python

To view a list all data streams, create a Provider instance and use its .list_streams() method:

from pprint import pprint

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# List all data streams.
streams = app.list_streams()
pprint(streams)

CLI

To view a list all data streams, use the imports:list command:

rai imports:list

To list data streams for a specific model, pass the model name to the --model flag:

rai imports:list --model MyModel

Get Data Stream Details

Requires the cdc_admin application role.

SQL

To get details about a specific data stream, pass the stream name and model name to the api.get_data_stream() procedure:

-- Get details about the data stream. Replace the placeholders with your
-- database, schema, and table or view name.
CALL relationalai.api.get_data_stream('<db>.<schema>.<table_or_view>');
/*+-----------------------------------------+-------------------------+-----------------------+--------+-------------------+--------------------------------------+-------------------------------+--------------+-------------------------------+------------------+------------------------+-------------------+-------------------------------+---------------------------------+-------------------------------+-------------------------------+----------------------------------------+--------+------------+
  | ID                                      | CREATED_AT              | CREATED_BY            | STATUS | REFERENCE_NAME    | REFERENCE_ALIAS                      | FQ_OBJECT_NAME                | RAI_DATABASE | RAI_RELATION                  | DATA_SYNC_STATUS | PENDING_BATCHES_COUNT  | NEXT_BATCH_STATUS | NEXT_BATCH_UNLOADED_TIMESTAMP | NEXT_BATCH_DETAILS              | LAST_BATCH_DETAILS            | LAST_BATCH_UNLOADED_TIMESTAMP | LAST_TRANSACTION_ID                    | ERRORS | CDC_STATUS |
  |-----------------------------------------+-------------------------+-----------------------+--------+-------------------+--------------------------------------+-------------------------------+--------------+-------------------------------+------------------+------------------------+-------------------+-------------------------------+---------------------------------+-------------------------------+-------------------------------+----------------------------------------+--------+------------|
  | ds_abcd1234_ef56_7890_abcd_1234ef567890 | 2024-10-23 10:12:34.567 | jane.doe@example.com  | ACTIVE | DATA_STREAM_TABLE | a1bcdef2-3456-7890-1234-b567c890d123 | <db>.<schema>.<table_or_view> | MyModel      | <db>.<schema>.<table_or_view> | SYNCED           | 0                      | NULL              | NULL                          | NULL                            | {"rows": 10, "size": 512, ... } 2024-10-23 10:50:00.456       | 02a1b234-5678-1234-abcdef-0123456789ab | []     | STARTED    |
  +-----------------------------------------+-------------------------+-----------------------+--------+-------------------+--------------------------------------+-------------------------------+--------------+-------------------------------+------------------+------------------------+-------------------+-------------------------------+---------------------------------+-------------------------------+-------------------------------+----------------------------------------+--------+------------+ */

Python

To get details about a specific data stream, create a Provider instance and use its .sql() method to execute the api.get_data_stream() SQL procedure with the name of the data stream and model:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# Get details about a data stream. Replace the placeholders with your database,
# schema, and table or view name.
stream_name = '<db>.<schema>.<table_or_view>'
stream_details = app.sql(f"CALL relationalai.api.get_data_stream({stream_name})")
print(stream_details)

CLI

To get details about a specific data stream, pass the stream ID to the --id option of the imports:get command:

rai imports:get --id <stream_id>

The stream ID is the unique identifier for the data stream, as returned in the output of the imports:list command.

View Quarantined Streams

Requires the cdc_admin application role.

SQL

To view quarantined streams and their errors, use the api.data_stream_batches view:

SELECT
  data_stream_id,
  fq_object_name,
  status,
  error.VALUE::string AS processing_error
FROM
  relationalai.api.data_stream_batches,
  LATERAL FLATTEN(input => processing_details:processingErrors) AS error
WHERE
  status = 'QUARANTINED';

Python

To view quarantined streams and their errors, create a Provider instance and use its .sql() method to query the api.data_stream_batches view:

import relationalai as rai

# Get a Provider instance.
app = rai.Provider()

# View quarantined streams and their errors.
quarantined_streams = app.sql("""
SELECT
  data_stream_id,
  fq_object_name,
  status,
  error.VALUE::string AS processing_error
FROM
    relationalai.api.data_stream_batches,
    LATERAL FLATTEN(input => processing_details:processingErrors) AS error
WHERE
    status = 'QUARANTINED';
""")

print(quarantined_streams)

Once the errors have been corrected, resume the data stream to continue processing. Note that quarantined streams continue to consume change tracking data from the source data, but these changes will not be processed by the CDC engine until the stream is resumed.