Apache Kafka (beta)
This an extension that receives and publishes events from/to Apache Kafka.
Features
A Kafka sink publishes events processed by GDN stream worker to a topic with a partition for a Kafka cluster. If the topic is not already created in the Kafka cluster, the Kafka sink creates the default partition for the given topic. The publishing topic and partition can be a dynamic value taken from the GDN stream worker event.
This sink is used to request replay of specific range of events on a specified partition of a topic.
A Kafka source receives events to be processed by GDN stream worker from a topic with a partition for a Kafka cluster. The events received can be in the
TEXTorJSONformat. If the topic is not already created in the Kafka cluster, the Kafka sink creates the default partition for the given topic.Kafka Replay Response (Source)
This source is used to listen to replayed events requested from kafka-replay-request sink
Sink
Kafka Sink
A Kafka sink publishes events processed by GDN stream worker to a topic with a partition for a Kafka cluster.If the topic is not already created in the Kafka cluster, the Kafka sink creates the default partition for the given topic. The publishing topic and partition can be a dynamic value taken from the GDN stream worker event.
To configure a sink to use the Kafka transport, the type parameter should have kafka as its value.
Syntax:
CREATE SINK <name> WITH (type="kafka", bootstrap.servers="<STRING>", topic="<STRING>", partition.no="<INT>", sequence.id="<STRING>", key="<STRING>", is.binary.message="<BOOL>", optional.configuration="<STRING>", map.type="<STRING>"...)
QUERY PARAMETERS:
| Name | Description | Default Value | Possible Data Types | Optional |
|---|---|---|---|---|
| bootstrap.servers | This parameter specifies the list of Kafka servers to which the Kafka sink must publish events. This list should be provided as a set of comma separated values. e.g., localhost:9092,localhost:9093. | STRING | No | |
| topic | The topic to which the Kafka sink needs to publish events. Only one topic must be specified. | STRING | No | |
| partition.no | The partition number for the given topic. Only one partition ID can be defined. If no value is specified for this parameter, the Kafka sink publishes to the default partition of the topic (i.e., 0). | 0 | INT | YES |
| sequence.id | A unique identifier to identify the messages published by this sink. This ID allows receivers to identify the sink that published a specific message. | null | STRING | YES |
| key | The key contains the values that are used to maintain ordering in a Kafka partition. | null | STRING | YES |
| is.binary.message | In order to send the binary events via kafka sink, this parameter is set to 'True'. | null | BOOL | NO |
| optional.configuration | This parameter contains all the other possible configurations that the producer is created with. e.g., producer.type:async,batch.size:200. | null | STRING | YES |
EXAMPLE 1:
@App:name('TestExecutionPlan')
CREATE STREAM FooStream (symbol string, price float, volume long);
@info(name = 'query1')
CREATE SINK BarStream WITH (type='kafka', topic='topic_with_partitions', partition.no='0', bootstrap.servers='localhost:9092', map.type='json') (symbol string, price float, volume long);
insert into BarStream
select symbol, price, volume
from FooStream;
This Kafka sink configuration publishes to 0th partition of the topic named topic_with_partitions.
EXAMPLE 2:
@App:name('TestExecutionPlan')
CREATE STREAM FooStream (symbol string, price float, volume long);
@info(name = 'query1')
CREATE SINK BarStream WITH (type='kafka', topic='{{symbol}}', partition.no='{{volume}}', bootstrap.servers='localhost:9092', map.type='json')) (symbol string, price float, volume long);
insert into BarStream
select symbol, price, volume
from FooStream ;
This query publishes dynamic topic and partitions that are taken from the GDN stream worker event. The value for partition.no is taken from the volume attribute, and the topic value is taken from the symbol attribute.
Kafka Replay Request
This sink is used to request replay of specific range of events on a specified partition of a topic.
Syntax:
CREATE SINK <name> WITH (type="kafka-replay-request", sink.id="<STRING>", @map.type="<STRING>")
QUERY PARAMETERS:
| Name | Description | Default Value | Possible Data Types | Optional | Dynamic |
|---|---|---|---|---|---|
| sink.id | a unique SINK_ID should be set. This sink id will be used to match with the appropriate kafka-replay-response source. | STRING | No | No |
EXAMPLE 1:
@App:name('TestKafkaReplay')
CREATE SINK BarStream WITH (type='kafka-replay-request', sink.id='1') (topicForReplay string, partitionForReplay string, startOffset string, endOffset string);
@info(name = 'query1')
CREATE SOURCE FooStream WITH (type='kafka-replay-response', group.id='group', threading.option='single.thread', bootstrap.servers='localhost:9092', sink.id='1', map.type='json') (symbol string, amount double);
CREATE SINK logStream (type='log') (symbol string, amount double);
insert into logStream
select *
from FooStream;
In this app we can send replay request events into BarStream and observe the replayed events in the logStream.
Source
Kafka Source
A Kafka source receives events to be processed by GDN stream worker from a topic with a partition for a Kafka cluster. The events received can be in the TEXT or JSON format. If the topic is not already created in the Kafka cluster, the Kafka sink creates the default partition for the given topic.
Syntax:
CREATE SOURCE <name> WITH (type="kafka", bootstrap.servers="<STRING>", topic.list="<STRING>", group.id="<STRING>", threading.option="<STRING>", partition.no.list="<STRING>", seq.enabled="<BOOL>", is.binary.message="<BOOL>", topic.offsets.map="<STRING>", enable.offsets.commit="<BOOL>", enable.async.commit="<BOOL>", optional.configuration="<STRING>", map.type="<STRING>")
QUERY PARAMETERS:
| Name | Description | Default Value | Possible Data Types | Optional |
|---|---|---|---|---|
| bootstrap.servers | This parameter specifies the list of Kafka servers to which the Kafka sink must publish events. This list should be provided as a set of comma separated values. e.g., localhost:9092,localhost:9093. | STRING | No | |
| topic.list | This specifies the list of topics to which the source must listen. This list can be provided as a set of comma-separated values. e.g., topic_one,topic_two | STRING | No | |
| group.id | This is an ID to identify the Kafka source group. The group ID ensures that sources with the same topic and partition that are in the same group do not receive the same event. | STRING | No | |
| threading.option | This specifies whether the Kafka source is to be run on a single thread, or in multiple threads based on a condition. Possible values are as follows: single.thread: To run the Kafka source on a single thread. topic.wise: To use a separate thread per topic. partition.wise: To use a separate thread per partition. | STRING | NO | |
| partition.no.list | The partition number list for the given topic. This is provided as a list of comma-separated values. e.g., 0,1,2,. | 0 | STRING | YES |
| seq.enabled | If this parameter is set to true, the sequence of the events received via the source is taken into account. Therefore, each event should contain a sequence number as an attribute value to indicate the sequence. | false | BOOL | YES |
| is.binary.message | In order to receive binary events via the Kafka source, it is required to set this parameter to 'True'. | false | BOOL | YES |
| is.binary.message | In order to receive binary events via the Kafka source, it is required to set this parameter to 'True'. | false | BOOL | YES |
| topic.offsets.map | This parameter specifies reading offsets for each topic and partition. The value for this parameter is specified in the following format: <topic>=<offset>,<topic>=<offset>, When an offset is defined for a topic, the Kafka source skips reading the message with the number specified as the offset as well as all the messages sent previous to that message. If the offset is not defined for a specific topic it reads messages from the beginning. e.g., stocks=100,trades=50 reads from the 101th message of the stocks topic, and from the 51st message of the trades topic. | null | STRING | YES |
| enable.offsets.commit | This parameter specifies whether to commit offsets. If the manual asynchronous offset committing is needed, enable.offsets.commit should be true and enable.auto.commit should be false. If periodical committing is needed enable.offsets.commit should be true and enable.auto.commit should be true. If committing is not needed, enable.offsets.commit should be false. Note: enable.auto.commit is an optional.configuration property. If it is set to true, Source will periodically(default: 1000ms. Configurable with auto.commit.interval.ms property as an optional.configuration) commit its current offset (defined as the offset of the next message to be read) for the partitions it is reading from back to Kafka. | true | BOOL | YES |
| enable.async.commit | This parameter will changes the type of the committing offsets returned on the last poll for the subscribed list of topics and partitions. When enable.async.commit is set to true, committing will be an asynchronous call. | true | BOOL | YES |
| optional.configuration | This parameter contains all the other possible configurations that the consumer is created with. e.g., ssl.keystore.type:JKS,batch.size:200. | null | STRING | YES |
EXAMPLE 1:
@App:name('TestExecutionPlan')
CREATE STREAM BarStream (symbol string, price float, volume long);
@info(name = 'query1')
CREATE SOURCE FooStream WITH (type='kafka', topic.list='kafka_topic,kafka_topic2', group.id='test', threading.option='partition.wise', bootstrap.servers='localhost:9092', partition.no.list='0,1', map.type='json') (symbol string, price float, volume long);
insert into BarStream
select symbol, price, volume
from FooStream;
This kafka source configuration listens to the kafka_topic and kafka_topic2 topics with 0 and 1 partitions. A thread is created for each topic and partition combination. The events are received in the json format, mapped to a GDN stream worker event, and sent to a stream named FooStream.
EXAMPLE 2:
@App:name('TestExecutionPlan')
CREATE STREAM BarStream (symbol string, price float, volume long);
@info(name = 'query1')
CREATE SOURCE FooStream WITH (type='kafka', topic.list='kafka_topic', group.id='test', threading.option='single.thread', bootstrap.servers='localhost:9092', map.type='json') (symbol string, price float, volume long);
insert into BarStream
select symbol, price, volume
from FooStream;
This Kafka source configuration listens to the kafka_topic topic for the default partition because no partition.no.list is defined. Only one thread is created for the topic. The events are received in the JSON format, mapped to a GDN stream worker event, and sent to a stream named FooStream.
EXAMPLE 3:
@App:name('TestExecutionPlan')
CREATE SOURCE FooStream WITH (type='kafka', topic.list='trp_topic', partition.no.list='0', threading.option='single.thread', group.id='group', bootstrap.servers='localhost:9092', map.type='json', enclosing.element='//events', map.attributes="symbol ='symbol', price = 'price', volume = 'volume', partition = 'trp:partition', topic = 'trp:topic', key = 'trp:key', recordTimestamp = 'trp:record.timestamp', eventTimestamp = 'trp:event.timestamp', checkSum = 'trp:check.sum', topicOffset = 'trp:offset'") (symbol string, price float, volume long, partition string, topic string, key string, recordTimestamp string, eventTimestamp string, checkSum string, topicOffset string);
insert into BarStream
select *
from FooStream;
This Kafka source configuration listens to the trp_topic topic for the default partition because no partition.no.list is defined.
Since the custom attribute mapping is enabled with TRP values, the GDN stream worker event will be populated with the relevant trp values as well.
Kafka Replay Request
This source is used to listen to replayed events requested from kafka-replay-request sink.
Syntax:
CREATE SOURCE <name> WITH (type="kafka-replay-response", bootstrap.servers="<STRING>", group.id="<STRING>", threading.option="<STRING>", sink.id="<INT>", map.type="<STRING>")
QUERY PARAMETERS:
| Name | Description | Default Value | Possible Data Types | Optional |
|---|---|---|---|---|
| bootstrap.servers | This parameter specifies the list of Kafka servers to which the Kafka sink must publish events. This list should be provided as a set of comma separated values. e.g., localhost:9092,localhost:9093. | STRING | No | |
| topic.list | This specifies the list of topics to which the source must listen. This list can be provided as a set of comma-separated values. e.g., topic_one,topic_two | STRING | No | |
| group.id | This is an ID to identify the Kafka source group. The group ID ensures that sources with the same topic and partition that are in the same group do not receive the same event. | STRING | No | |
| threading.option | This specifies whether the Kafka source is to be run on a single thread, or in multiple threads based on a condition. Possible values are as follows: single.thread: To run the Kafka source on a single thread. topic.wise: To use a separate thread per topic. partition.wise: To use a separate thread per partition. | STRING | NO | |
| sink.id | a unique SINK_ID . | INT | No |
EXAMPLE 1:
@App:name('TestKafkaReplay')
CREATE SINK BarStream WITH (type='kafka-replay-request', sink.id='1') (topicForReplay string, partitionForReplay string, startOffset string, endOffset string);
@info(name = 'query1')
CREATE SOURCE FooStream WITH (type='kafka-replay-response', group.id='group', threading.option='single.thread', bootstrap.servers='localhost:9092', sink.id='1', map.type='json') (symbol string, amount double);
CREATE SINK logStream WITH (type='log') (symbol string, amount double);
insert into logStream
select *
from FooStream;
In this app we can send replay request events into BarStream and observe the replayed events in the logStream.