Overview
Numeric Typesβ
Doris supports the following numeric data types:
BOOLEANβ
There are two possible values: 0 represents false, and 1 represents true.
For more info, please refer BOOLEANγ
Integerβ
All are signed integers. The differences among the INT types are the number of bytes occupied and the range of values they can represent:
-
TINYINT: 1 byte, [-128, 127]
-
SMALLINT: 2 bytes, [-32768, 32767]
-
INT: 4 bytes, [-2147483648, 2147483647]
-
BIGINT: 8 bytes, [-9223372036854775808, 9223372036854775807]
-
LARGEINT: 16 bytes, [-2^127, 2^127 - 1]
Floating-pointβ
Including imprecise floating-point types FLOAT and DOUBLE, corresponding to the float
and double
in common programming languages
Fixed-pointβ
The precise fixed-point type DECIMAL, used in financial and other cases that require strict accuracy.
Date Typesβ
Date types include DATE, TIME and DATETIME, DATE type only stores the date accurate to the day, DATETIME type stores the date and time, which can be accurate to microseconds. TIME type only stores the time, and does not support the construction of the table storage for the time being, can only be used in the query process.
Do calculation for datetime types or converting them to numeric types, please use functions like TIME_TO_SEC, DATE_DIFF, UNIX_TIMESTAMP . The result of directly converting them as numeric types as not guaranteed.
For more information refer to DATE, TIME and DATETIME documents.
String Typesβ
Doris supports both fixed-length and variable-length strings, including:
-
CHAR(M): A fixed-length string, where M is the byte length. The range for M is [1, 255].
-
VARCHAR(M): A variable-length string, where M is the maximum length. The range for M is [1, 65533].
-
STRING: A variable-length string with a default maximum length of 1,048,576 bytes (1 MB). This maximum length can be increased up to 2,147,483,643 bytes (2 GB) by configuring the
string_type_length_soft_limit_bytes
setting.
Semi-Structured Typesβ
Doris supports different semi-structured data types for JSON data processing, each tailored to different use cases.
-
ARRAY / MAP / STRUCT: They support nested data and fixed schema, making them well-suited for analytical workloads such as user behavior and profile analysis, as well as querying data lake formats like Parquet. Due to the fixed schema, there is no overhead for dynamic schema inference, resulting in high write and analysis performance.
-
VARIANT: It supports nested data and flexible schema. It is well-suited for analytical workloads such as log, trace, and IoT data analysis. It can accommodate any legal JSON data, which will be automatically expanded into sub-columns in a columnar storage format. This approach enables high compression rate in storage and high performance in data aggregation, filtering, and sorting.
-
JSON: It supports nested data and flexible schema. It is optimized for high-concurrency point query use cases. The flexible schema allows for ingesting any legal JSON data, which will be stored in a binary format. Extracting fields from this binary JSON format is more than 2X faster than using regular JSON strings.
Aggregation Typesβ
The aggregation data types store aggregation results or intermediate results during aggregation. They are used for accelerating aggregation-heavy queries.
-
BITMAP: It is used for exact deduplication, such as in (UV) statistics and audience segmentation. It works in conjunction with BITMAP functions like
bitmap_union
,bitmap_union_count
,bitmap_hash
, andbitmap_hash64
. -
HLL: It is used for approximate deduplication and provides better performance than
COUNT DISTINCT
. It works in conjunction with HLL functions likehll_union_agg
,hll_raw_agg
,hll_cardinality
, andhll_hash
. -
QUANTILE_STATE: It is used for approximate percentile calculations and offers better performance than the
PERCENTILE
function. It works with functions likeQUANTILE_PERCENT
,QUANTILE_UNION
, andTO_QUANTILE_STATE
. -
AGG_STATE: It is used to accelerate aggregations, utilized in combination with aggregation function combinators like state/merge/union.
IP Typesβ
IP data types store IP addresses in a binary format, which is faster and more space-efficient for querying compared to storing them as strings. There are two supported IP data types: