Data Export

About 9 min

Data Export

1. Function Overview

IoTDB supports three methods for data export:

Data Export Tool: export-data.sh/bat is located in the tools directory. It can export the query results of specified SQL statements into CSV, SQL, and TsFile (open-source time-series file format) files.
PIPE Framework-based TsFileBackup: tsfile-backup.sh/bat is located in the tools directory. It can export specified data files into TsFile format using the PIPE framework.
Copy SQL Export to TsFile: Writes query results back to a TsFile at the specified path via SQL.

File Format	IoTDB Tool	Description
CSV	export-data.sh/bat	Plain text format for storing structured data. Must follow the CSV format specified below.
SQL		File containing custom SQL statements.
TsFile		Open-source time-series file format.
	tsfile-backup.sh/bat	An open-source time-series data file format,and this script supports the Object data type.
	Copy SQL	Open-source time-series file format.

2. Data Export Tool

2.1 Common Parameters

Short	Full Parameter	Description	Required	Default
`-ft`	`--file_type`	Export file type: `csv`, `sql`, `tsfile`.	Yes	-
`-h`	`--host`	Hostname of the IoTDB server.	No	`127.0.0.1`
`-p`	`--port`	Port number of the IoTDB server.	No	`6667`
`-u`	`--username`	Username for authentication.	No	`root`
`-pw`	`--password`	Password for authentication. Supported for hidden input since V2.0.9.1	No	`TimechoDB@2021`(Before V2.0.6 it is root)
`-sql_dialect`	`--sql_dialect`	Select server model : tree or table	No	tree
`-db`	`--database`	The target database to be exported only takes effect when `-sql_dialect` is of the table type.	Yes when `-sql_dialect = table`	-
`-table`	`--table`	The target table to be exported only takes effect when `-sql_dialect` is of the table type. If the `-q` parameter is specified, this parameter will not take effect. If the export type is tsfile/sql, this parameter is mandatory.	No	-
`-start_time`	`--start_time`	The start time of the data to be exported only takes effect when `-sql_dialect` is of the table type. If `-q` is specified, this parameter will not take effect. The supported time formats are the same as those for the `-tf` parameter.	No	-
`-end_time`	`--end_time`	The end time of the data to be exported only takes effect when `-sql_dialect` is set to the table type. If `-q` is specified, this parameter will not take effect.	No	-
`-t`	`--target`	Target directory for the output files. If the path does not exist, it will be created.	Yes	-
`-pfn`	`--prefix_file_name`	Prefix for the exported file names. For example, `abc` will generate files like `abc_0.tsfile`, `abc_1.tsfile`.	No	`dump_0.tsfile`
`-q`	`--query`	SQL query command to execute. Starting from v2.0.8, semicolons in SQL statements are automatically removed, and query execution proceeds normally.	No	-
`-timeout`	`--query_timeout`	Query timeout in milliseconds (ms).	No	`-1` (before v2.0.8) `Long.MAX_VALUE` (v2.0.8 and later) (Range: `-1~Long.MAX_VALUE`)
`-help`	`--help`	Display help information.	No	-
`-usessl`	`--use_ssl`	Use SSL protocol. Supported since V2.0.9.1	No	-
`-ts`	`--trust_store`	Trust store. Supports hidden input. Supported since V2.0.9.1	No	-
`-tpw`	`--trust_store_password`	Trust store password. Supports hidden input. Supported since V2.0.9.1	No	-

2.2 CSV Format

2.2.1 Command

# Unix/OS X
> tools/export-data.sh -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>] 
                -t <target_directory> [-pfn <prefix_file_name>] [-dt <datatype>] [-lpf <lines_per_file>] [-tf <time_format>] 
               [-tz <timezone>] [-q <query_command>] [-timeout <query_timeout>]
# Windows
# Before version V2.0.4.x
> tools\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>] 
                -t <target_directory> [-pfn <prefix_file_name>] [-dt <datatype>] [-lpf <lines_per_file>] [-tf <time_format>] 
               [-tz <timezone>] [-q <query_command>] [-timeout <query_timeout>]
               
# V2.0.4.x and later versions
> tools\windows\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>] 
                -t <target_directory> [-pfn <prefix_file_name>] [-dt <datatype>] [-lpf <lines_per_file>] [-tf <time_format>] 
               [-tz <timezone>] [-q <query_command>] [-timeout <query_timeout>]

2.2.2 CSV-Specific Parameters

Short	Full Parameter	Description	Required	Default
`-dt`	`--datatype`	Whether to include data types in the CSV file header (`true` or `false`).	No	`false`
`-lpf`	`--lines_per_file`	Number of rows per exported file.	No	`10000` (Range：0～Integer.Max=2147483647)
`-tf`	`--time_format`	Time format for the CSV file. Options: 1) Timestamp (numeric, long), 2) ISO8601 (default), 3) Custom pattern (e.g., `yyyy-MM-dd HH:mm:ss`). SQL file timestamps are unaffected by this setting.	No	`ISO8601`
`-tz`	`--timezone`	Timezone setting (e.g., `+08:00`, `-01:00`).	No	System default

2.2.3 Examples

# Valid Example
> export-data.sh -ft csv -sql_dialect table -t /path/export/dir -db database1 -q "select * from table1"

# Error Example
> export-data.sh -ft csv -sql_dialect table -t /path/export/dir -q "select * from table1"
Parse error: Missing required option: db

2.3 SQL Format

2.3.1 Command

# Unix/OS X
> tools/export-data.sh -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>] 
          -t <target_directory> [-pfn <prefix_file_name>] [-aligned <export aligned insert sql>]
          -lpf <lines_per_file> - [-tf <time_format>] [-tz <timezone>] [-q <query_command>] [-timeout <query_timeout>]
      
# Windows
# Before version V2.0.4.x 
> tools\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host> -p <port> -u <username> -pw <password>] 
          -t <target_directory> [-pfn <prefix_file_name>  -aligned <export aligned insert sql>  
          -lpf <lines_per_file> -tf <time_format> -tz <timezone> -q <query_command> -timeout <query_timeout>]
          
# V2.0.4.x and later versions
> tools\windows\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host> -p <port> -u <username> -pw <password>] 
          -t <target_directory> [-pfn <prefix_file_name>  -aligned <export aligned insert sql>  
          -lpf <lines_per_file> -tf <time_format> -tz <timezone> -q <query_command> -timeout <query_timeout>]

2.3.2 SQL-Specific Parameters

Short	Full Parameter	Description	Required	Default
`-aligned`	`--use_aligned`	Whether to export as aligned SQL format (`true` or `false`).	No	`true`
`-lpf`	`--lines_per_file`	Number of rows per exported file.	No	`10000` (Range：0～Integer.Max=2147483647)
`-tf`	`--time_format`	Time format for the CSV file. Options: 1) Timestamp (numeric, long), 2) ISO8601 (default), 3) Custom pattern (e.g., `yyyy-MM-dd HH:mm:ss`). SQL file timestamps are unaffected by this setting.	No	`ISO8601`
`-tz`	`--timezone`	Timezone setting (e.g., `+08:00`, `-01:00`).	No	System default

2.3.3 Examples

# Valid Example
> export-data.sh -ft sql -sql_dialect table -t /path/export/dir -db database1 -start_time 1

# Error Example
> export-data.sh -ft sql -sql_dialect table -t /path/export/dir -start_time 1
Parse error: Missing required option: db

2.4 TsFile Format

2.4.1 Command

# Unix/OS X
> tools/export-data.sh -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>]
        -t <target_directory> [-pfn <prefix_file_name>] [-q <query_command>] [-timeout <query_timeout>]
      
# Windows
# Before version V2.0.4.x 
> tools\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>]
        -t <target_directory> [-pfn <prefix_file_name>] [-q <query_command>] [-timeout <query_timeout>]
        
# V2.0.4.x and later versions
> tools\windows\export-data.bat -ft<format> [-sql_dialect<sql_dialect>] -db<database> -table<table>  
                [-start_time<start_time>] [-end_time<end_time>] [-h <host>] [-p <port>] [-u <username>] [-pw <password>]
        -t <target_directory> [-pfn <prefix_file_name>] [-q <query_command>] [-timeout <query_timeout>]

2.4.2 TsFile-Specific Parameters

None

2.4.3 Examples

# Valid Example
> /tools/export-data.sh -ft tsfile -sql_dialect table -t /path/export/dir -db database1 -start_time 0

# Error Example
> /tools/export-data.sh -ft tsfile -sql_dialect table -t /path/export/dir -start_time 0
Parse error: Missing required option: db

3. TsFileBackup Based on PIPE Framework

Since V2.0.9.2, IoTDB supports the tsfile-backup.sh/bat script. This script can automatically generate and send the CREATE PIPE SQL command to the server, exporting specified data files to TsFile format.

Notes:

To use this script, contact the Timecho Team to obtain the JAR package(tsfile-remote-sink-<version>-jar-with-dependencies.jar), and place it in a path accessible to IoTDB (e.g., all Data Node hosts).
This script supports exporting Object-type data to TsFile files.

3.1 Execution Commands

# Unix/OS X
> tools/tsfile-backup.sh [-sql_dialect <sql_dialect>] [-h <host>] [-p <port>]
       [-u <username>] [-pw <password>] [-path <path>] [-db <db>] [-table
       <table>] [-s <start_time>] [-e <end_time>] [-t <target_directory>] 
       [-th <target_host>] [-tu <target_host_user>] [-tp <target_host_port>]
       [--rate_limit] [--plugin_jar] [-help]

# Windows
> tools\windows\tsfile-backup.bat [-sql_dialect <sql_dialect>] [-h <host>] [-p <port>]
       [-u <username>] [-pw <password>] [-path <path>]  [-db <db>] [-table
       <table>] [-s <start_time>] [-e <end_time>] [-t <target_directory>] 
       [-th <target_host>] [-tu <target_host_user>] [-tp <target_host_port>]
       [--rate_limit] [--plugin_jar] [-help]

3.2 Script Parameters

Abbreviation	Full Name	Description	Required	Default
`-sql_dialect`	`--sql_dialect`	Specifies the data model type. Valid values: `tree` (Tree Model) or `table` (Table Model).	Yes	-
`-h`	`--host`	Local host address (IP of the IoTDB instance where the data resides).	No	`127.0.0.1`
`-p`	`--port`	Port number for the IoTDB RPC service.	No	`6667`
`-u`	`--user`	Username for IoTDB authentication.	No	`root`
`-pw`	`--password`	Password for IoTDB authentication (hidden input supported).	No	`root`
`-t`	`--target`	Export target directory. In SCP mode, this is an absolute physical path on the remote server. TsFile and associated Object directories will be exported here.	Yes	-
`-db`	`--database`	Database name (optional for Table Model).	No	`.*`
`-table`	`--table`	Table name (optional for Table Model).	No	`.*`
`-s`	`--start_time`	Start time (ISO8601 format e.g. `2026-01-01T00:00:00` or millisecond timestamp). Only data from this time onwards is exported.	No	-
`-e`	`--end_time`	End time (same format as above). Only data before this time is exported.	No	-
`-th`	`--target_host`	Remote target host IP. If specified, the script automatically configures Pipe to use SCP for data transfer.	No	-
`-tu`	`--target_host_user`	Username for SSH/SCP login to the remote server.	No	-
`-tpw`	`--target_host_pw`	Password for remote authentication (hidden input supported).	No	-
`-tp`	`--target_host_port`	Remote SSH port.	No	`22`
`--rate_limit`	`--rate_limit`	Transfer rate limit (unit: Bytes/s) to prevent excessive bandwidth usage.	No	-
`--plugin_jar`	`--plugin_jar`	Path to the Pipe plugin JAR file.	No	-
`--object-parallelism`	`--object-parallelism`	Specifies the maximum parallelism for object file transmission.	No	-
`--object-batch-size`	`--object-batch-size`	Limits the total byte size of each object file upload batch, used to control memory usage and single SCP transfer size.	No	-
`-help`	`--help`	Show help information.	No	-

3.3 Execution Examples

Example 1: SCP Remote Export (Send Data to Another Server)

./tsfile-backup.sh -sql_dialect table -db test_db -t /remote/archive/ -th 192.168.1.100 -tu backup_user -tpw ComplexPass123!

Example 2: Remote Object Data Export with Rate Limiting

./tsfile-backup.sh -sql_dialect table -t /mnt/backup/ -th 10.0.0.5 -tu iot_admin -tpw Admin@2026 --rate_limit 5242880

Example 3: Specify Pipe Plugin JAR Directory

./tsfile-backup.sh -sql_dialect table -db test -table .* -tu luoluoyuyu -tpw -t /tmp/backup --plugin_jar /local/lib/tsfile-remote-sink-2.0.8-SNAPSHOT-jar-with-dependencies.jar

Note: When exporting Object-type data in SCP mode, to avoid handshake exceptions, connection failures, or frequent Pipe restarts, it is recommended to take any of the following measures:

Appropriately lower the configuration parameter object-parallelism
Increase the MaxStartups value on the target machine as needed. After modification, execute sshd reload or sshd restart for the configuration to take effect.

4. Copy SQL

Note: This feature is supported since V2.0.11.1.

4.1 Command

// ---------------------------------------- Copy Statement ---------------------------------------------------------
copyToStatement
    : COPY '(' query ')' TO fileName=string ((WITH)? copyToStatementOptions)?
    | COPY tableName=qualifiedName ('(' tableColumns=identifierList ')')? TO fileName=string ((WITH)? copyToStatementOptions)?
    ;

copyToStatementOptions
    : '(' copyToStatementOption (',' copyToStatementOption)* ')'
    ;

copyToStatementOption
    : FORMAT identifier
    | TABLE identifier
    | TAGS '(' identifierList ')'
    | TIME identifier
    | MEMORY_THRESHOLD memory=INTEGER_VALUE
    ;

Parameters

Name	Description	Default
FORMAT	Export format. Currently only `TsFile` is supported.	TsFile
TABLE	Specifies the table name in the generated TsFile.	If the query SQL involves only one table, that table name is used; otherwise `default` is used.
TIME	Specifies which column in the result set is used as the TIME column. When manually specified: an error is reported if the column type is not TIMESTAMP or the specified column cannot be found. When not manually specified: an error is also reported if a column named `time` exists but its type is not TIMESTAMP. The time column is constructed with the following priority: 1. A column with the same name as the time column of the single table involved in the query. 2. A column named "time" with type TIMESTAMP is used as the time column. 3. The current row count written by the corresponding device is used as the time to generate the time column, with the column name "time".	-
TAGS	Specifies which columns are TAG columns. When there are multiple TAG columns, the order in the final generated table is consistent with the specified order. When manually specified: an error is reported if the column does not exist, the column type is not STRING, or duplicate column names exist. If the query involves only one table and all tag columns of that table can be found in the query result set, they are inferred as the tag columns of that table; otherwise, the default value is an empty list, meaning all columns except the TIME column are treated as FIELD columns.	Empty list
MEMORY_THRESHOLD	Used for memory control when generating TsFile (unit: byte). An error is reported when the manually specified value is less than or equal to 0.	32MB

Result Set

Column	Data Type	Description
path	STRING	The absolute path of the generated target file
row_count	INT64	Total number of rows written
device_count	INT64	Number of devices generated
size_in_bytes	INT64	Size of the generated target file
table_name	STRING	The table name in the target file. If it is auto-generated, it is marked with `(auto_gen)`.
time_column	STRING	The name of the time column in the table of the target file. If it is auto-generated, it is marked with `(auto_gen)`.
tag_columns	STRING	The names of the tag columns in the table of the target file, separated by `,`.

Other Notes

File generation location:
- If a file name is specified, the generated TsFile is saved under ${dn_data_dirs}/copy_to of the DataNode directly connected to the client. When multiple directories are configured, the file is generated according to the strategy specified by the configuration item dn_multi_dir_strategy.
- If a path is specified, the file is saved under the specified path.
Possible exceptions during execution:
- An error is reported when out-of-order timestamps exist while writing to TsFile according to the given schema.
- An error is reported for illegal file names or when the target file already exists.
- Duplicate column names exist in the query result.
- Insufficient disk space.

4.2 Examples

Taking table1 in the Sample Data as an example.

Export all data from table1 to the file copysql1.tsfile via a SELECT statement.

TimechoDB:database1> copy (select * from table1) to 'copysql1.tsfile'
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
|                                                 path|row_count|device_count|size_in_bytes|table_name|time_column|                  tag_columns|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
|/timechodb/data/datanode/data/copy_to/copysql1.tsfile|       18|           6|         4636|    table1|       time|[region, plant_id, device_id]|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
Total line number = 1
It costs 0.336s

Export all data from table1 to the file copysql2.tsfile via the table name.

TimechoDB:database1> copy table1 to 'copysql2.tsfile'
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
|                                                 path|row_count|device_count|size_in_bytes|table_name|time_column|                  tag_columns|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
|/timechodb/data/datanode/data/copy_to/copysql2.tsfile|       18|           6|         4636|    table1|       time|[region, plant_id, device_id]|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-----------------------------+
Total line number = 1
It costs 0.048s

Export partial data from table1 to the file copysql3.tsfile via table name (columns).

TimechoDB:database1> copy table1 (device_id,temperature) to 'copysql3.tsfile'
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
|                                                 path|row_count|device_count|size_in_bytes|table_name|   time_column|tag_columns|
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
|/timechodb/data/datanode/data/copy_to/copysql3.tsfile|       18|           1|          558|    table1|time(auto_gen)|         []|
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
Total line number = 1
It costs 0.064s

Export the aggregation result of partial data from table1 to the file copysql4.tsfile via a SELECT statement.

TimechoDB:database1> copy (select count(temperature), count(humidity) from table1 group by device_id) to 'copysql4.tsfile'
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
|                                                 path|row_count|device_count|size_in_bytes|table_name|   time_column|tag_columns|
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
|/timechodb/data/datanode/data/copy_to/copysql4.tsfile|        2|           1|          543|    table1|time(auto_gen)|         []|
+-----------------------------------------------------+---------+------------+-------------+----------+--------------+-----------+
Total line number = 1
It costs 0.155s

Export partial data from table1 to the file copysql5.tsfile via a SELECT statement, with the target table, time column, and tag columns specified.

TimechoDB:database1> copy (select time,region,device_id,temperature from table1 order by time) to 'copysql5.tsfile' (TABLE copytable, TIME time, TAGS (region,device_id))
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-------------------+
|                                                 path|row_count|device_count|size_in_bytes|table_name|time_column|        tag_columns|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-------------------+
|/timechodb/data/datanode/data/copy_to/copysql5.tsfile|       18|           4|         1199| copytable|       time|[region, device_id]|
+-----------------------------------------------------+---------+------------+-------------+----------+-----------+-------------------+
Total line number = 1
It costs 0.047s