Source Type | Examples | Description |
---|---|---|
REST |
|
In the pipeline, the Ingest REST API passes REST data to Kafka. REST sources use a REST API key, which is generated when you create the source, to ingest data and send messages. |
MQTT |
|
An MQTT source uses a central broker with messages organized by topics. You can subscribe to one or more MQTT topics on an MQTT broker. In the data pipeline, Listener writes messages from the MQTT Subscriber to source topics in Kafka. You can optionally secure the MQTT subscription using an SSL certificate and private key. Messages within MQTT have an associated QoS (Quality of Service) level, which determines the broker's level of effort to ensure a message is received. Listener supports messages with QoS level 0 (send at most once) or QoS level 1 (send at least once).
|
- Ingests the REST and MQTT data streams into Kafka.
Kafka uses a write-ahead-log buffer to store and manage data sent to it.
- Writers configured for a REST or MQTT source start reading the REST and MQTT data at specific batch intervals.
- The writer associated with a specific target system destination writes the REST or MQTT data streams to the target system destination.
Writers manage how REST and MQTT data is stored, or persisted in the following target system destinations:
- Teradata Database
Teradata targets can be configured to use a JDBC driver or Teradata QueryGrid to write data. Teradata QueryGrid can distribute data to Teradata Database systems to achieve high throughput when the data ingestion rate is high.
- HDFS
HDFS and Hbase targets write data in sequence file format to a specified directory.
- Hbase
HDFS and Hbase targets write data in sequence file format to a specified directory.
- Aster
Aster targets use JDBC to write data.
- Broadcast streams
Listener sends broadcast streams to external apps through a WebSocket server.
For more information about writers, see How Listener Writes to Targets.
- Teradata Database
In Listener, the REST API manages sources, users, and targets, and provides Listener status.
- Uses Kafka to store data so service interruptions in target systems do not result in data loss. However, by default, Kafka holds data for only 72 hours. If the data is not consumed within 72 hours, the data is lost. The 72-hour default is configurable.
- Shows trends in data flow from sources and detects variances that may indicate problems upstream.
Listener supports up to 10,000 active systems, 10,000 active sources, and 10,000 active targets.