Dremio Cloud
Dremio Software
 
On this page

    Data Types

    A data type is a classification of data that determines the type of value a variable possesses and the types of mathematical, relational, or logical operations that can be performed on it. Dremio supports the following SQL data types: numeric, string and binary, boolean, date and time, and semi-structured.

    The following topics related to data types are covered:

    Summary of Supported Data Types in Dremio

    Dremio supports the following SQL data types.

    CategoryData TypeNotesExamples
    Numeric Data TypesDECIMALA DECIMAL type has precision (p) and scale (s): DECIMAL(p,s). Precision is the total number of digits. Scale is the number of digits to the right of the decimal point. When you perform arithmetic on DECIMAL types that have different precision and/or scale, the return value will have sufficient precision and scale to hold the result of the operation. The following are decimal limitations in Dremio:

    • Decimal numeric literals in SQL queries cannot be larger than the maximum BIGINT value, which is 9223372036854775807.

    • Queries that perform arithmetic operations on a column and literal may cause errors. For example, SELECT CAST(12345 as DOUBLE) * CAST(A as DOUBLE) fails. The workaround is to use a string value such as SELECT CAST('12345' as DOUBLE) * CAST(A as DOUBLE).

    • Queries casting numeric literals to decimal should use specific precision. Precision of literal cannot be lowered. For example, CAST(123.23 as DECIMAL(2,0)) returns the same number as the output since the given number cannot be represented using a precision of 2.

    • When there is an overflow with the decimal arithmetic output, the returned result will overflow.

    		987.65 is a DECIMAL(5, 2) value
    INTA 4-byte signed integer. The supported range is from -2147483648 to 2147483647.5135
    BIGINTAn 8-byte signed integer. The supported range is from -9223372036854775808 to 9223372036854775807.-749826542587
    FLOATA 4-byte single-precision floating point. A FLOAT provides six decimal digits of precision.123.123456
    DOUBLE8-byte double-precision floating point. A DOUBLE provides 15 decimal digits of precision.123.123456789012345
    String & Binary Data TypesVARCHARVARCHAR stands for variable-length character string. By default, the maximum allowed length is is 32,000 bytes. VARCHAR supports only UTF-8 encoded values.18852367854
    VARBINARYVARBINARY stands for variable-length binary string. By default, the maximum allowed length is 32,000 bytes. The value must be entered as a string value.SELECT CAST ('help' as VARBINARY)

    -- aGVscA==

    Boolean Data TypeBOOLEANThe supported values for BOOLEAN include true, false, and null.TRUE, FALSE, and NULL
    Date & Time Data TypesDATEA date value that enables you to calculate and store consistent information about the date of the events and transactions.

    Note:

    When using a string literal for the date, yyyy-mm-dd is the only supported format. To use a different format, use the TO_DATE() function.

    		DATE ‘2000-01-01’
    TIMEIdentifies the time of day, which enables you to calculate and store consistent information about the time of the events and transactions.

    Note:

    When using a string literal for the time, HH24:MI:SS.sss and HH24:MI:SS are the only supported formats. To use a different format, use the TO_TIME() function.

    		TIME ‘17:30:50.235’

    TIME ‘17:30:50’

    TIMESTAMPRepresents an absolute point in time with millisecond precision without a time zone. Timestamps are truncated to the nearest millisecond. For more information, see Time Zone Support.
    • TIMESTAMP ‘2000-01-01 01:30:50’
    • TIMESTAMP ‘2000-01-01 17:30:50’
    • TIMESTAMP ‘2000-01-01 17:30:50.9’
    • TIMESTAMP ‘2000-01-01 17:30:50.12’
    • TIMESTAMP ‘2000-01-01 17:30:50.123’
    INTERVAL (day to seconds)
    INTERVAL (years to months)    		Intervals are used to represent a measure of time. Dremio supports the two available types of intervals: year-month, which stores the year and month (YYYY-MM); and day-time (DD HH:MM:SS), which stores the days, hours, minutes, and seconds.

    Additionally, the following forms are supported:

    • DAY HOUR:MINUTE:SECOND:MILLISECOND - For example, INTERVAL '3' DAY
    • YEAR-MONTH - For example, INTERVAL '3' MONTH
    • YEAR-MONTH - For example, INTERVAL '1' YEAR
    • DAY - For example, INTERVAL '5' DAY
    • MINUTE - For example, INTERVAL '5' MINUTE
    • SECOND - For example, INTERVAL '5' SECOND
    • DAY TO HOUR - For example, INTERVAL '4 01' DAY TO HOUR
    • DAY TO MINUTE - For example, INTERVAL '4 01:01' DAY TO MINUTE
    • DAY TO SECOND - For example, INTERVAL '4 01:01:01' DAY TO SECOND
    • INTERVAL ‘1 2:34:56.789’ DAY TO SECOND
    • INTERVAL ‘1-5’ YEAR TO MONTH
    Semi-structured Data TypesSTRUCTUsed to represent collections of key-value pairs. Keys are non-empty, case-insensitive strings, and values can be of any type. The example shows the required format for a query where the key (city) must be enclosed in [ ] and the column (address) is a STRUCT data type.

    Note:

    Dremio does not have STRUCT literals, but you can get the same result using CONVERT_FROM and JSON strings. For example:

    SELECT CONVERT_FROM('{"name":"Gnarly", "age":7, "car":null}', 'json')

    -- {"name:"Gnarly","age":7}

    		SELECT address['city'] FROM customerTable.
    LISTUsed to represent a list of arbitrary size, where the index is a non-negative integer and values can be of any type. The elements in a LIST must be of the same type (for example, a list consisting of only integers). The example shows the required format for a query where the index (100) must be enclosed in [ ] and the column (OrderHistoryTable) is a LIST data type.

    Note:

    Dremio does not have LIST literals, but you can get the same result using CONVERT_FROM and JSON strings. For example:

    SELECT CONVERT_FROM('["apple", "strawberry", "banana"]', 'json')

    -- ["apple","strawberry","banana"]

    		SELECT customerOrders[100] FROM OrderHistoryTable
    MAPThe MAP type is a collection of key-value pairs. MAP keys are case-insensitive strings. All values in a given map have the same type. For example, map<string, int> represents a mapping where the keys are strings and the values are integers. To retrieve the value of a MAP element, use column['key'] syntax: SELECT <column_name['<key_name>']> FROM <table_name>. For information about the SQL functions that are available for MAP expressions, see Datatype.

    Note:

    If you have queried tables with MAP data using an earlier release of Dremio, you must run ALTER TABLE table_name FORGET METADATA on those tables so that Dremio knows they have MAP rather than STRUCT. Otherwise, Dremio will give an error prompting you to reformat your dataset.

    		SELECT address['city'] FROM customerTable