Kafka to StarRocks

BladePipe supports data replication from Kafka to StarRocks. View supported migration, sync, verification, and connector capabilities.

Target DataSource：

StarRocks

Connection

Function	Description
Incremental Data Sync	Allow subscribing to messages from the source Topic and transforming them into DML operations, namely, INSERT, UPDATE, DELETE.
Subscription Modification	Add, delete, or modify the subscribed topics. For more information, see Modify Subscription.
Position Resetting	Reset positions by timestamp to consume the data in a past period again.

Function	Description
Message Format	The following message formats are supported. See Message Format. CloudCanal format AlibabaCanal format
Removal of Target Data before Full Data Migration	Remove the existing data in the Target before running the Full Data Migration, applicable for DataJobs reruning and scheduled Full Data migrations.
Recreating Target Table	Recreate target tables before running the Full Data Migration, applicable for DataJobs reruning and scheduled Full Data migrations.
Stream Load	Use Stream Load to write data to StarRocks BE. By default, batch write is adopted, with dynamic adjustment of data flush interval and batch size.
Handling of Zero Value for Time	Allow setting zero value for time to different data types to prevent errors when writing to the Target.

Limit	Description
Creating Tables in the Target in Advance	Only support automatic Topic creation for messages.
Raw Message Format	Only support raw message replication from Kafka to Kafka, and Raw Message Format needs to be selected at both the Source and the Target.
Target Table Type	Only support Primary Key model.
Source Table Type	Migration and sync of tables without primary keys are not supported.
DDL Synchronization Errors	Continuous DDLs on the same table will cause errors (because asynchronous DDLs are executed on a target StarRocks instance). Errors may occur when modifying field constraints or some types of DDL. If DDL errors occur, you can change the target table schema and then skip the errors by setting DataJob parameters.
Incremental Data Write Conflict Resolution Rule	Using Stream Load method, the primary key is used for full row replacement.

Prerequisite	Description
Port Preparation	Allow the migration and sync node (Worker) to connect to the Kafka ports.

Parameter	Description
schemaFormat	MQ Message format. For more information, see Message Format.
consumerGroupId	Kafka consumer group ID.
consumeParallel	Degree of consuming Kafka topics in parallel.
sessionTimeoutMs	Kafka session timeout in milliseconds.
maxPollRecords	Maximum number of messages fetched in one poll from Kafka.
dbHeartbeatIntervalSec	Interval for initiating heartbeat on the source database.
dbHeartbeatToleranceStep	The threshold of gap between the latest offset and the current offset. If the actual gap is bigger than the threshold, BladePipe won‘t send heartbeat message.
customClientProps	Custom properties passed to the Kafka client in JSON format. The key is the parameter name and the value is the parameter value. This setting takes the highest priority. For example: AWS IAM Access Control

Tips: To modify the general parameters, see General Parameters and Functions.

Prerequisite	Description
Permissions for Account	SELECT and DDL permissions (optional)
Port Preparation	Allow the migration and sync node (Worker) to connect to the StarRocks FE QueryPort and FE/BE HttpPort.

Parameter	Description
host	MySQL port, corresponding to StarRocks FE QueryPort.
httpHost	Host for StarRocks stream load, corresponding to StarRocks FE/BE HttpPort.
totalDataInMemMb	Maximum data size allowed in memory when writing in batches; If the data size exceeds the memory limit, or the wait time exceeds asyncFlushIntervalSec, then data is flushed to the write queue.
asyncFlushIntervalSec	Interval to wait for flushing when writing in batches; If the wait time exceeds asyncFlushIntervalSec, or the data size exceeds totalDataInMemMb, then data is flushed to the write queue.
flushBatchMb	Maximum batch size per table; If the batch size exceeds this limit, then data is flushed to the write queue.
realFlushPauseSec	Wait time to flush data to StarRocks using stream load, 0 means no wait is needed.
soTimeoutSec	TCP socket timeout (so_timeout) during QueryPort operations.
httpSoTimeoutSec	TCP socket timeout (so_timeout) during HttpPort operations.
enableTimeZoneProcess	Enable time zone conversion for time fields.
timezone	Timezone in the Target, e.g., +08:00 Asia/Shanghai America/New_York.
maxInSizePerQuery	Maximum number of IN clause values per query during secondary verification. Queries exceeding this limit will be automatically split.

Tips: To modify the general parameters, see General Parameters and Functions.