Druid connector

The Druid connector allows querying an Apache Druid database from Trino.

Requirements

To connect to Druid, you need:

• Druid version 0.22.0 or higher.
• Network access from the Trino coordinator and workers to your Druid broker. Port 8082 is the default port.

Configuration

Create a catalog properties file that specifies the Druid connector by setting the connector.name to druid and configuring the connection-url with the JDBC string to connect to Druid.

For example, to access a database as example, create the file etc/catalog/example.properties. Replace BROKER:8082 with the correct host and port of your Druid broker.

connector.name=druid
connection-url=jdbc:avatica:remote:url=http://BROKER:8082/druid/v2/sql/avatica/

You can add authentication details to connect to a Druid deployment that is secured by basic authentication by updating the URL and adding credentials:

connection-url=jdbc:avatica:remote:url=http://BROKER:port/druid/v2/sql/avatica/;authentication=BASIC
connection-user=root
connection-password=secret

Now you can access your Druid database in Trino with the example catalog name from the properties file.

The connection-user and connection-password are typically required and determine the user credentials for the connection, often a service user. You can use secrets to avoid actual values in the catalog properties files.

Additionally, the following configuration properties can be set depending on the use-case:

Druid Configuration Properties

Property Name	Description	Default
druid.execution-timeout	Query timeout for Druid queries, beyond which unfinished queries will be cancelled. Example: 10s, 1m, and 1000ms.	None

Data source authentication

The connector can provide credentials for the data source connection in multiple ways:

• inline, in the connector configuration file
• in a separate properties file
• in a key store file
• as extra credentials set when connecting to Trino

You can use secrets to avoid storing sensitive values in the catalog properties files.

The following table describes configuration properties for connection credentials:

Property name	Description
credential-provider.type	Type of the credential provider. Must be one of INLINE, FILE, or KEYSTORE; defaults to INLINE.
connection-user	Connection user name.
connection-password	Connection password.
user-credential-name	Name of the extra credentials property, whose value to use as the user name. See extraCredentials in Parameter reference.
password-credential-name	Name of the extra credentials property, whose value to use as the password.
connection-credential-file	Location of the properties file where credentials are present. It must contain the connection-user and connection-password properties.
keystore-file-path	The location of the Java Keystore file, from which to read credentials.
keystore-type	File format of the keystore file, for example JKS or PEM.
keystore-password	Password for the key store.
keystore-user-credential-name	Name of the key store entity to use as the user name.
keystore-user-credential-password	Password for the user name key store entity.
keystore-password-credential-name	Name of the key store entity to use as the password.
keystore-password-credential-password	Password for the password key store entity.

General configuration properties

The following table describes general catalog configuration properties for the connector:

Property name	Description
case-insensitive-name-matching	Support case insensitive schema and table names. Defaults to false.
case-insensitive-name-matching.cache-ttl	Duration for which case insensitive schema and table names are cached. Defaults to 1m.
case-insensitive-name-matching.config-file	Path to a name mapping configuration file in JSON format that allows Trino to disambiguate between schemas and tables with similar names in different cases. Defaults to null.
case-insensitive-name-matching.config-file.refresh-period	Frequency with which Trino checks the name matching configuration file for changes. The duration value defaults to 0s (refresh disabled).
metadata.cache-ttl	Duration for which metadata, including table and column statistics, is cached. Defaults to 0s (caching disabled).
metadata.cache-missing	Cache the fact that metadata, including table and column statistics, is not available. Defaults to false.
metadata.schemas.cache-ttl	Duration for which schema metadata is cached. Defaults to the value of metadata.cache-ttl.
metadata.tables.cache-ttl	Duration for which table metadata is cached. Defaults to the value of metadata.cache-ttl.
metadata.statistics.cache-ttl	Duration for which tables statistics are cached. Defaults to the value of metadata.cache-ttl.
metadata.cache-maximum-size	Maximum number of objects stored in the metadata cache. Defaults to 10000.
write.batch-size	Maximum number of statements in a batched execution. Do not change this setting from the default. Non-default values may negatively impact performance. Defaults to 1000.
dynamic-filtering.enabled	Push down dynamic filters into JDBC queries. Defaults to true.
dynamic-filtering.wait-timeout	Maximum duration for which Trino waits for dynamic filters to be collected from the build side of joins before starting a JDBC query. Using a large timeout can potentially result in more detailed dynamic filters. However, it can also increase latency for some queries. Defaults to 20s.

Appending query metadata

The optional parameter query.comment-format allows you to configure a SQL comment that is sent to the datasource with each query. The format of this comment can contain any characters and the following metadata:

• $QUERY_ID: The identifier of the query.
• $USER: The name of the user who submits the query to Trino.
• $SOURCE: The identifier of the client tool used to submit the query, for example trino-cli.
• $TRACE_TOKEN: The trace token configured with the client tool.

The comment can provide more context about the query. This additional information is available in the logs of the datasource. To include environment variables from the Trino cluster with the comment , use the ${ENV:VARIABLE-NAME} syntax.

The following example sets a simple comment that identifies each query sent by Trino:

query.comment-format=Query sent by Trino.

With this configuration, a query such as SELECT * FROM example_table; is sent to the datasource with the comment appended:

SELECT * FROM example_table; /*Query sent by Trino.*/

The following example improves on the preceding example by using metadata:

query.comment-format=Query $QUERY_ID sent by user $USER from Trino.

If Jane sent the query with the query identifier 20230622_180528_00000_bkizg, the following comment string is sent to the datasource:

SELECT * FROM example_table; /*Query 20230622_180528_00000_bkizg sent by user Jane from Trino.*/

note

Certain JDBC driver settings and logging configurations might cause the comment to be removed.

Domain compaction threshold

Pushing down a large list of predicates to the data source can compromise performance. Trino compacts large predicates into a simpler range predicate by default to ensure a balance between performance and predicate pushdown. If necessary, the threshold for this compaction can be increased to improve performance when the data source is capable of taking advantage of large predicates. Increasing this threshold may improve pushdown of large dynamic filters. The domain-compaction-threshold catalog configuration property or the domain_compaction_threshold catalog session property can be used to adjust the default value of {{default_domain_compaction_threshold}} for this threshold.

Case insensitive matching

When case-insensitive-name-matching is set to true, Trino is able to query non-lowercase schemas and tables by maintaining a mapping of the lowercase name to the actual name in the remote system. However, if two schemas and/or tables have names that differ only in case (such as "customers" and "Customers") then Trino fails to query them due to ambiguity.

In these cases, use the case-insensitive-name-matching.config-file catalog configuration property to specify a configuration file that maps these remote schemas and tables to their respective Trino schemas and tables. Additionally, the JSON file must include both the schemas and tables properties, even if only as empty arrays.

{
  "schemas": [
    {
      "remoteSchema": "CaseSensitiveName",
      "mapping": "case_insensitive_1"
    },
    {
      "remoteSchema": "cASEsENSITIVEnAME",
      "mapping": "case_insensitive_2"
    }],
  "tables": [
    {
      "remoteSchema": "CaseSensitiveName",
      "remoteTable": "tablex",
      "mapping": "table_1"
    },
    {
      "remoteSchema": "CaseSensitiveName",
      "remoteTable": "TABLEX",
      "mapping": "table_2"
    }]
}

Queries against one of the tables or schemes defined in the mapping attributes are run against the corresponding remote entity. For example, a query against tables in the case_insensitive_1 schema is forwarded to the CaseSensitiveName schema and a query against case_insensitive_2 is forwarded to the cASEsENSITIVEnAME schema.

At the table mapping level, a query on case_insensitive_1.table_1 as configured above is forwarded to CaseSensitiveName.tablex, and a query on case_insensitive_1.table_2 is forwarded to CaseSensitiveName.TABLEX.

By default, when a change is made to the mapping configuration file, Trino must be restarted to load the changes. Optionally, you can set the case-insensitive-name-matching.config-file.refresh-period to have Trino refresh the properties without requiring a restart:

case-insensitive-name-matching.config-file.refresh-period=30s

Type mapping

Because Trino and Druid each support types that the other does not, this connector modifies some types when reading data.

Druid type to Trino type mapping

The connector maps Druid types to the corresponding Trino types according to the following table:

Druid type to Trino type mapping

Druid type	Trino type	Notes
STRING	VARCHAR
FLOAT	REAL
DOUBLE	DOUBLE
LONG	BIGINT	Except for the special _time column, which is mapped to TIMESTAMP.
TIMESTAMP	TIMESTAMP	Only applicable to the special _time column.

No other data types are supported.

For results that match Trino semantics, Druid must use SQL-compatible null handling (druid.generic.useDefaultValueForNull=false), which preserves NULL values for all data types. This is the default since Druid 28.0.0 and mandatory from Druid 32.0.0; on earlier versions set it explicitly. With Druid's legacy null handling (druid.generic.useDefaultValueForNull=true), NULL is replaced with default values such as 0 for LONG, 0.0 for DOUBLE, and an empty string '' for STRING. Those results do not match Trino semantics, including for pushed-down aggregations.

Type mapping configuration properties

The following properties can be used to configure how data types from the connected data source are mapped to Trino data types and how the metadata is cached in Trino.

Property name	Description	Default value
unsupported-type-handling	Configure how unsupported column data types are handled: * IGNORE, column is not accessible. * CONVERT_TO_VARCHAR, column is converted to unbounded VARCHAR. The respective catalog session property is unsupported_type_handling.	IGNORE
jdbc-types-mapped-to-varchar	Allow forced mapping of comma separated lists of data types to convert to unbounded VARCHAR

SQL support

The connector provides read access to data and metadata in the Druid database. In addition to the globally available and read operation statements, the connector supports the following features:

• Druid connector
• Druid connector

Procedures

system.flush_metadata_cache()

Flush JDBC metadata caches. For example, the following system call flushes the metadata caches for all schemas in the example catalog

USE example.example_schema;
CALL system.flush_metadata_cache();

system.execute('query')

The execute procedure allows you to execute a query in the underlying data source directly. The query must use supported syntax of the connected data source. Use the procedure to access features which are not available in Trino or to execute queries that return no result set and therefore can not be used with the query or raw_query pass-through table function. Typical use cases are statements that create or alter objects, and require native feature such as constraints, default values, automatic identifier creation, or indexes. Queries can also invoke statements that insert, update, or delete data, and do not return any data as a result.

The query text is not parsed by Trino, only passed through, and therefore only subject to any security or access control of the underlying data source.

The following example sets the current database to the example_schema of the example catalog. Then it calls the procedure in that schema to drop the default value from your_column on your_table table using the standard SQL syntax in the parameter value assigned for query:

USE example.example_schema;
CALL system.execute(query => 'ALTER TABLE your_table ALTER COLUMN your_column DROP DEFAULT');

Verify that the specific database supports this syntax, and adapt as necessary based on the documentation for the specific connected database and database version.

Table functions

The connector provides specific table functions to access Druid.

query(varchar) -> table

The query function allows you to query the underlying database directly. It requires syntax native to Druid, because the full query is pushed down and processed in Druid. This can be useful for accessing native features which are not available in Trino or for improving query performance in situations where running a query natively may be faster.

The native query passed to the underlying data source is required to return a table as a result set. Only the data source performs validation or security checks for these queries using its own configuration. Trino does not perform these tasks. Only use passthrough queries to read data.

As an example, query the example catalog and use STRING_TO_MV and MV_LENGTH from Druid SQL's multi-value string functions to split and then count the number of comma-separated values in a column:

SELECT
  num_reports
FROM
  TABLE(
    example.system.query(
      query => 'SELECT
        MV_LENGTH(
          STRING_TO_MV(direct_reports, ",")
        ) AS num_reports
      FROM company.managers'
    )
  );

note

The query engine does not preserve the order of the results of this function. If the passed query contains an ORDER BY clause, the function result may not be ordered as expected.

Performance

Pushdown

The connector supports aggregate pushdown for the following functions:

• avg()
• count()
• max()
• min()
• sum()

Pushed-down aggregations follow Trino's NULL semantics, which relies on Druid's SQL-compatible null handling (druid.generic.useDefaultValueForNull=false, the default since Druid 28.0.0). For example, avg, min, max, and sum over empty input return NULL rather than Druid's legacy default values.

If your Druid cluster must keep legacy null handling enabled (druid.generic.useDefaultValueForNull=true), disable aggregate pushdown so that aggregation results stay consistent with Trino's NULL semantics. Set the aggregation-pushdown.enabled catalog configuration property (or the aggregation_pushdown_enabled catalog session property) to false.

note

The connector performs pushdown where performance may be improved, but in order to preserve correctness an operation may not be pushed down. When pushdown of an operation may result in better performance but risks correctness, the connector prioritizes correctness.

Predicate pushdown support

The connector supports predicate expression pushdown. Comparisons between numeric columns (=, <>, <, <=, >, >=) are pushed down to Druid rather than evaluated by Trino, including predicates that cannot be expressed as a range, such as comparisons between two columns. Arithmetic operators and LIKE expressions are not pushed down.