Documentation Plugins Snowflake data source for Grafana

Snowflake data source for Grafana

The Snowflake data source plugin allows you to query and visualize Snowflake data metrics from within Grafana.

Requirements

This plugin has the following requirements:

• A Snowflake user with the appropriate role granted.
  • This data source does not require a specific role.
  • The Snowflake user’s role is what allows that user to access tables. In order to query your data, ensure your user has the appropriate roles.
• One of the following account types:
• Any free or paid Grafana Cloud plan or an activated on-prem Grafana Enterprise license. Contracted Cloud customers should refer to their agreement.

Install the Snowflake data source plugin

To install the data source, refer to Installation

Configure Snowflake

Configuring the Snowflake data source requires a Snowflake user with a username and a password.

Grafana recommends creating a new user with limited permissions for this data source.

Create a user

In order to connect to Snowflake, you must create a user or authenticate using an existing one. This user will run all queries sent from Grafana.

If you would like to have separate users run different queries / workloads, then you should create multiple Snowflake data sources with different settings.

To create a user in Snowflake, you will have to log in to your Snowflake instance and run the CREATE USER command.

Grant a role

Now that the Snowflake user is created, the user must be granted a role using the GRANT ROLE command. Granting a role to a user allows the user to perform operations allowed by that role.

This role is what defines what warehouses and tables the user has access to.

Configure the data source in Grafana

These connection settings are the same that are used in connecting via SnowSQL

Add a data source by filling in the following fields:

Key Pair authentication

For enhanced security, Key Pair authentication can be utilized as an alternative to basic authentication. You can generate the public and private keys by following the referenced Snowflake documentation. When using Key Pair authentication, it is important to update the rsa_public_key in Snowflake and provide the username and unencrypted private key in the data source configuration.

OAuth authentication

You can use OAuth authentication to pass through tokens to Snowflake on behalf of the user logged into Grafana. Below provides some instructions on using Azure AD as the OAuth provider.

1. Use Azure AD to setup OAuth

2. Follow instructions here to update the application you created in step 1, and add a client application for Snowflake.

3. Update the scopes you created in step 1 in your grafana.ini file. Add the api you created in step 2. scopes should look something like:

Copy

scopes = api://8c1a0b1c-6bb0-4190-a730-8a1c34237619/session:role-any openid email profile offline_access

4. Restart Grafana and Login with Azure AD. Create a Snowflake Datasource using Authentication Type: OAuth and toggle Forward OAuth Identity. Save and test to confirm the token is being passed through and is valid.

What if I get an invalid token error? Step 2 instructions provide ways to validate the token which will provide additional information on why it is invalid.

Copy

select system$verify_external_oauth_token('<ACCESS_TOKEN>');

Additional Links:

Azure OAuth. Follow the “on behalf of user” Flow: https://docs.snowflake.com/en/user-guide/oauth-azure Testing with Postman: https://community.snowflake.com/s/article/How-To-Configure-Postman-for-testing-SQL-API-with-OAuth Setup a Snowflake security integration: https://community.snowflake.com/s/article/Create-Security-Integration-User-To-Use-With-OAuth-Client-Token-With-Azure-AD

Configure the data source with provisioning

It is possible to configure data sources using config files with Grafana’s provisioning system. You can read more about how it works and all the settings you can set for data sources on the provisioning docs page

Example:

yaml Copy

datasources:
  - name: Snowflake
    type: grafana-snowflake-datasource
    access: proxy
    basicAuth: false
    editable: true
    enabled: true
    jsonData:
      account: xyz123.east-us-2.azure
      username: grafana-user
      authType: password
      timeInterval: 10s
      defaultQuery: SELECT \n\t $__timeGroup(<time_column>, $__interval) as time,\n\t <value_column>\n FROM <metric_table>\n WHERE $__timeFilter(time)
      defaultVariableQuery: SELECT DISTINCT <column_name> FROM <metric_table> LIMIT 1000
      defaultInterpolation: '' # Refer: https://grafana.com/docs/grafana/latest/variables/advanced-variable-format-options/
    secureJsonData:
      password: grafana-password
  - name: Snowflake Billing Data
    type: grafana-snowflake-datasource
    access: proxy
    basicAuth: false
    editable: true
    enabled: true
    jsonData:
      account: xyz123.us-east1.gcp
      username: grafana-admin-user
      database: snowflake
      role: ACCOUNTADMIN
      timeInterval: 10s
      defaultQuery: SELECT \n\t $__timeGroup(<time_column>, $__interval) as time,\n\t <value_column>\n FROM <metric_table>\n WHERE $__timeFilter(time)
      defaultVariableQuery: SELECT DISTINCT <column_name> FROM <metric_table> LIMIT 1000
      defaultInterpolation: sqlstring # Refer: https://grafana.com/docs/grafana/latest/variables/advanced-variable-format-options/
    secureJsonData:
      password: grafana-admin-password

Query the data source

The provided query editor is a standard SQL query editor. Grafana includes some macros to help with writing more complex timeseries queries.

Macros

Examples

Table Visualization

Most queries in Snowflake will be best represented by a table visualization. Any query will display data in a table. If it can be queried, then it can be put in a table.

This example returns results for a table visualization:

sql Copy

SELECT {column_1}, {column_2} FROM {table};

Timeseries / Graph visualizations

For timeseries / graph visualizations, there are a few requirements:

• A column with a date or datetime type must be selected
• The date column must be in ascending order (using ORDER BY column ASC)
• A numeric column must also be selected

To make a more reasonable graph, be sure to use the $__timeFilter and $__timeGroup macros.

Example timeseries query:

sql Copy

SELECT
  avg(execution_time) AS average_execution_time,
  $__timeGroup(start_time, $__interval),
  query_type
FROM
  account_usage.query_history
WHERE
  $__timeFilter(start_time)
group by
  query_type,start_time
order by
  start_time,query_type ASC;

Inspecting the query

Because Grafana supports macros that Snowflake does not, the fully rendered query, which can be copy/pasted directly into Snowflake, is visible in the Query Inspector. To view the full interpolated query, click the Query Inspector button, and the full query will be visible under the “Query” tab. If you need to debug further check out the “Query History” page in Snowsight. Learn more here: https://docs.snowflake.com/en/user-guide/ui-snowsight-activity

Templates and variables

To add a new Snowflake query variable, refer to Add a query variable. Use your Snowflake data source as your data source for the following available queries:

Any value queried from a Snowflake table can be used as a variable. Be sure to avoid selecting too many values, as this can cause performance issues.

If the variable query returns two columns, the values from the second column will be used as display values

Using variables

After creating a variable, you can use it in your Snowflake queries by using Variable syntax. For more information about variables, refer to Templates and variables.

While using the variables in your query, optionally you can set the interpolation format. You can also configure the default interpolation format in the datasource config.

Single-value variables

If a variable returns a single value, you can use one of the following formats. The below example assumes, you have set sqlstring as default interpolation and you have two variables called queryTypeSingle and limit. They have values SELECT and 2 respectively.

sql Copy

SELECT query_type FROM account_usage.query_history WHERE query_type = ${queryTypeSingle} Limit ${limit:raw}

will be translated into

sql Copy

SELECT query_type FROM account_usage.query_history WHERE query_type = 'SELECT' Limit 10

Setting default variable interpolation type is introduced in version 1.2. From this version,you can set the default interpolation type to sqlstring

Prior to 1.2 version of the plugin or if you use none as the default interpolation type, the same query have to be written as the following

sql Copy

SELECT query_type FROM account_usage.query_history WHERE query_type = '${queryTypeSingle}' Limit ${limit:raw}

Multi-value variables using sqlstring interpolation

When consuming a variable which return multiple options, you can consume it in the following method. The below example assumes, you have set sqlstring as default interpolation and you have variables called queryTypeMulti and limit. They have values CREATE,SELECT and 2 respectively.

sql Copy

SELECT query_type FROM account_usage.query_history WHERE query_type in (${queryTypeMulti}) Limit ${limit:raw}

will be translated into

sql Copy

SELECT query_type FROM account_usage.query_history WHERE query_type in ('CREATE','SELECT') Limit 10

Multi-value variables using regex

To use a variable that has multiple values, you can use the regex modifier option and the regexp Snowflake function. E.g. ${variable:regex}

For example, this query will filter using only the Query Types selected in the queryType variable:

sql Copy

...
  AND query_type regexp '${queryType:regex}'
...

This query translates to:

sql Copy

...
  AND query_type regexp '(DESCRIBE|CREATE_USER|DROP|TRUNCATE_TABLE|ALTER)'
...

Visualizing data as Logs

With the Logs format selected in the query, you can visualize the data in Logs viewer in explorer. When querying with Logs format, your query should have at least one time column and one string/content column. Optionally, query can have third column called level to set the log level of the particular row. Supported log level and their keywords can be found in grafana logs integration docs site. If the query returns any additional columns, they will be treated as additional fields/ detected fields in the logs.

For example, following is the valid log query.

sql Copy

SELECT 'hello foo'  as "content", (timestamp '2021-12-31') as "start_time", 'warn'  as "level" UNION
SELECT 'hello bar'  as "content", (timestamp '2021-12-30 14:12:59') as "start_time", 'error' as "level" UNION
SELECT 'hello baz'  as "content", (timestamp '2021-12-30') as "start_time", 'warn'  as "level" UNION
SELECT 'hello qux'  as "content", (timestamp '2021-12-29') as "start_time", 'info'  as "level" UNION
SELECT 'hello world' as "content", (timestamp '2021-12-28') as "start_time", 'unknown'  as "level" UNION
SELECT 'hello user' as "content", (timestamp '2021-12-27') as "start_time", 'info'  as "level"

Import a dashboard for Snowflake

The bundled Snowflake dashboard requires a data source with an ACCOUNTADMIN role for viewing billing data. There is an example of this in the provisioning section.

Follow these instructions for importing a dashboard.

Imported dashboards can be found in Configuration > Data Sources > select your Snowflake data source > select the Dashboards tab to see available pre-made dashboards.

Get the most out of the plugin

• Add Annotations.
• Configure and use Templates and variables.
• Add Transformations.
• Use Logs in explorer.
• Trace frame data structure.
• Set up alerting; refer to Alerts overview.

Read More

• Key Pair Authentication
• Monitor query activity with Query History in Snowflake