Data Types
GreptimeDB supports the following data types:
| Type name | Description | Synonyms | Size |
|---|---|---|---|
tinyint | 8-bit signed small integers between -128~127 | 1 Byte | |
smallint | 16-bit signed big integers between -32768~32767 | 2 Bytes | |
int | 32-bit signed integers between -2147483648~2147483647 | integer | 4 Bytes |
bigint | 64-bit signed big integers between -9223372036854775808~9223372036854775807 | 8 Bytes | |
varchar | UTF-8 encoded strings | text/ string/ char | The length of the strings |
float | 32-bit IEEE754 floating point values | 4 Bytes | |
double | Double precision IEEE 754 floating point values | 8 Bytes | |
boolean | Boolean values | 1 Byte | |
varbinary | Variable length binary values | The length of the data + 2 bytes | |
date | 32-bit date values | 4 Bytes | |
datetime | 64-bit datetime values | 8 Bytes | |
timestamp[(0/3/6/9)] | 64-bit timestamp values with optional precision. For example, timestamp(0) represents timestamp type with seconds precision, timestamp(3) represents milliseconds precision, timestamp(6) for microseonds and timestamp(9) for nanoseconds. If no precision is given, the timestamp is in milliseconds precision by default. | 8 Bytes |
Unsigned version of integer types
int / tinyint / smallint / bigint also have unsigned version, and there corresponding value ranges are:
int unsigned: 0~4294967295tinyint unsigned: 0~255smallint unsigned: 0~65535bigint unsigned: 0~18446744073709551615
Variable-sized type limitations
The max capacities of variable-sized type, such as string and varbinary are determined by their encodings and how storage engine handles them.
For example, string values are encoded into UTF-8. If all characters are 3-bytes lengthed, this field can store 715827882 characters. As for varbinary types, it can store 2147483647 bytes at most.
Choose the data type for timestamp column
GreptimeDB allows user to choose bigint or timestamp for timestamp index column. Both bigint and timestamp are interpreted as timestamp in millisecond precision.
SQL
# using TIMESTAMP as timestamp column data type
CREATE TABLE monitor (
host STRING,
ts TIMESTAMP,
cpu DOUBLE DEFAULT 0,
memory DOUBLE,
TIME INDEX (ts),
PRIMARY KEY(host)) ENGINE=mito WITH(regions=1);
# using BIGINT as timestamp column data type is also allowed
CREATE TABLE monitor (
host STRING,
ts BIGINT,
cpu DOUBLE DEFAULT 0,
memory DOUBLE,
TIME INDEX (ts),
PRIMARY KEY(host)) ENGINE=mito WITH(regions=1);