Table SQL Statements
Tables can be created in either the shared
$scratch location or in filesystem sources types.
In addition, if tables are created in a source, they can also be altered and dropped.
The following SQL commands are documented on this page:
CREATE TABLE AS
CREATE TABLE AS command is used to create tables.
Tables are created in either the shared
$scratch location or in filesystem sources types.
Tables are written in Parquet format and can be used with
HASH PARTITION BY and
By default, the
$scratch location points to the
/scratch directory under Dremio’s
configured distributed cache location (
paths.dist property in dremio.conf).
There are no security protocols or permissions associated with this location – it is readable and writable by all users.
Dremio CTAS supports all filesystem source types (S3, ADLS, NAS, HDFS, and MapR-FS) using the filesystem permissions for the written table using impersonation.
Enabling CTAS and DROP TABLE for Sources
In order to use the
CREATE TABLE AS (CTAS) and
DROP TABLE (DROP) commands on a source,
you must add these privileges to the appropriate users for the source.
If enabled, all users/groups who have access to the source can leverage CTAS and DROP TABLE on sources.
Privileges can be enabled:
- When creating a new source.
- By editing an existing source.
To enable the appropriate privileges for the source:
- In the Users table, identify the user or group that you want to edit privileges for, and click the cell to enable or disable the
Dropprivilege for the user.
- Click Save.
CTAS functionality is enabled on a per source basis.
For sources that support impersonation (HDFS and MapR File System) and if impersonation is enabled, Dremio executes the CTAS and DROP TABLE operations as the user running the command. This requires that the user have the appropriate permissions (read, execute, and/or write) for the operation. If impersonation is not enabled, the operation is executed as the user running the Dremio service user.
For additional security, access can be further controlled by selecting specifically restricted filesystem sub-directories when you initially create your Dremio source.
- Virtual datasets can be exported to their respective sources via standard SQL-based commands ONLY.
- DROP TABLE functionality on S3 or ADLS source datasets is slow when working with very large tables.
Syntax$scratch location syntax
CREATE TABLE $scratch.<tableName> [HASH PARTITION BY (<columnName>, . .) ] [LOCALSORT BY (<columnName>) ] AS <QUERY>
CREATE TABLE <source>.<tableName> [HASH PARTITION BY (<columnName>, . .) ] [LOCALSORT BY (<columnName>) ] AS <QUERY>
ExamplesCreating a $scratch table
CREATE TABLE $scratch.my_table AS select * from TPCH.lineitem
CREATE TABLE s3.upload.lineitem2 AS select * from TPCH.lineitem
Once a source table has been created, the Dremio UI displays the following:
- Path where the table was created
- Number of records in the new table
- File size of the table
Querying TablesQuerying a $scratch table
select count(*) from $scratch.my_table
select count(*) from s3.upload.lineitem2
If you specify an alias for a column or an expression in the SELECT clause, you can refer to that alias elsewhere in the query, including in the SELECT list or in the WHERE clause.Example 1
SELECT c_custkey AS c, lower(c) FROM "customer.parquet"
SELECT c_custkey AS c, lower(c) FROM ( SELECT c_custkey, c_mktsegment AS c FROM "customer.parquet")
SELECT c_name AS n, n FROM ( SELECT c_mktsegment AS n, c_name FROM "customer.parquet") AS MY_TABLE WHERE n = 'BUILDING'
SELECT c_custkey FROM ( SELECT c_custkey, c_name AS c FROM "customer.parquet" ) WHERE c = 'aa'
SELECT * FROM ( SELECT c_custkey AS c, c_name FROM "customer.parquet" ) JOIN "orders.parquet" ON c = o_orderkey
SELECT c_custkey AS c FROM "customer.parquet" JOIN "orders.parquet" ON c = o_orderkey
DESCRIBE TABLE command is used to provide high-level information regarding the overall column properties of an existing dataset.
DESCRIBE supports all filesystem source types and relies upon a user’s existing privileges to access and describe a table or view.
DESCRIBE TABLE table_name;
DESCRIBE TABLE taxistats;
Once a table or view has been queried, the Dremio UI displays the following:
- Column headers indicating the type of information being described
- Rows describing each table column
- Data types, properties, policies, and other associated attributes
The cells containing a ‘’ indicate “empty” values.
Version Considerations: For companies using Dremio v16.0+, users must have the
ALTER privilege granted for a dataset or folder in order to use this command. For earlier versions of access control on Dremio, users must have the
Can Edit privilege to use this command.
Modifies the structure of columns in a table, such as adding, deleting, or modifying columns in an existing table. To create a table, refer to the
CREATE TABLE command.
ALTER TABLE <tableName> ADD COLUMNS (<columnName1> <dataType1>(<size1>), <columnName2> <dataType2>(<size2>), ...); ALTER TABLE <tableName> DROP COLUMN <columnName>; ALTER TABLE <tableName> ALTER COLUMN <sourceColumnName> <newColumnName> <newDataType>(<size>); ALTER TABLE <tableName> MODIFY COLUMN <sourceColumnName> <newColumnName> <newDataType>(<size>); ALTER TABLE <tableName> REFRESH METADATA; ALTER TABLE <tableName> REFRESH METADATA FOR PARTITIONS ( "<partitionName>" = '<value>' );
The 〈size〉 parameter is optional.
ADD COLUMNS command creates new column(s) with the specified name, data type, and character limit. See the example below for how to employ this.
ALTER TABLE customer1 ADD COLUMNS (SSN INT(9))
DROP COLUMN command deletes the specified column, along with all data stored in it. As this cannot be undone, ensure that you’ve secured a backup or snapshot of your databases. See the example below for how to employ this.
ALTER TABLE customer1 DROP COLUMN SSN
ALTER COLUMN and
MODIFY COLUMN commands are used to change the data type for specific column(s). This effectively alters how queries read the data stored in the column(s), so the current data type should be similar, if not identical, to the way the new data type is read.
Depending on your source, you will need to use either
ALTER COLUMN or
MODIFY COLUMN specifically. For example,SQL Server uses
ALTER COLUMN whereas Oracle uses
MODIFY COLUMN. For additional information, see the ALTER TABLE - ALTER/MODIFY COLUMN resource.
See the example below for how to employ this.Altering a column
ALTER TABLE customer1 ALTER COLUMN SSN CHAR(11)
About ALTER/MODIFY Data Types: We recommend against changing data types for a column if it already has records on it. Unless you’ve gone through the column to ensure no issues are found with the data format, this may create issues regarding how the data is read. The likely end result is myriad
NULL values. Where possible, it is safer to add a new column with the desired data type and then drop the column whose data you wish to read differently.
Version Considerations: This SQL command may only be used with Dremio 18.0+.
ALTER TABLE command may be used to manually refresh the metadata associated with a table. This should be coupled with Improved Metadata Refreshes to fully benefit from any improved functionality.
You may perform metadata refreshes one of two ways: full or partial.
The example below shows how to run a full metadata refresh.Refreshing metadata
ALTER TABLE customer1 REFRESH METADATA
In comparison, the example below demonstrates a metadata refresh for a single partition.Refreshing metadata for a single partition
ALTER TABLE customer1 REFRESH METADATA FOR PARTITIONS ( "dir0" = 'a')
This example identifies the level of the partition to use (
"dir0") and the specific partition (
'a') to use for the refresh.
ALTER TABLE command consists of the following parameters for which values must be included:
The name of the table being altered must be included in the command. This helps to restrict the scope of the change rather than accidentally apply a
DROP COLUMNS on every dataset in Dremio. This must match the name of your desired table.
Dremio cannot perform any alteration to column data without the column itself specified. This helps to refine and restrict the scope of the changes.
This specifies the type of data to be stored for records if a new column is being added or an existing column being modified.
(size) component of this parameter defines the limitation on the total number of allowed characters that may be stored on each row for this column.
For a list of data types and their maximum allowed sizes, see SQL Data Types.
Tables can be dropped only if they have been created on a filesystem source and if you have permission.Syntax
DROP TABLE <source>.<tableName>
DROP TABLE s3.upload.lineitem2