[Demo+Webinar] New Product Updates to Make Serverless Flink a Developer’s Best Friend | Watch Now

The Simplest Useful Kafka Connect Data Pipeline in the World…or Thereabouts – Part 2

Written By

In the previous article in this blog series I showed how easy it is to stream data out of a database into Apache Kafka®, using the Kafka Connect API. I used MySQL in my example, but it’s equally applicable to any other database that supports JDBC—which is pretty much all of them! Now we’ll take a look at how we can stream data, such as that brought in from a database, out of Kafka and into Elasticsearch.

Kafka Connect

Again, we’re using Kafka Connect to do this in a scalable, fault-tolerant way, and all with just some simple configuration files!

Here I’m assuming you’ve followed the previous article for the general setup and installation steps. If you haven’t already, start Elasticsearch:

$ elasticsearch
[...]
[2017-08-01T15:22:40,132][INFO ][o.e.h.n.Netty4HttpServerTransport] [-1Kfx0p] publish_address {127.0.0.1:9201}, bound_addresses {[fe80::1]:9201}, {[::1]:9201}, {127.0.0.1:9201}, {127.94.0.2:9201}, {127.94.0.1:9201}
[2017-08-01T15:22:40,132][INFO ][o.e.n.Node               ] [-1Kfx0p] started

Now with further ado, let’s dive in!

Whilst Kafka Connect is part of Apache Kafka itself, if you want to stream data from Kafka to Elasticsearch you’ll want the Confluent Platform (or at least, the Elasticsearch connector).

The configuration is pretty simple. As before, see inline comments for details

It’s worth noting that if you’re using the same convertor throughout your pipelines (Avro, in this case) you’d actually put this in the Connect worker config itself rather than repeating it for each connector configuration.

Load the connector using the Confluent CLI:

./bin/confluent load es-sink-mysql-foobar-01 -d /tmp/kafka-connect-elasticsearch-sink.json

As with the File sink above, as soon as the connector is created and starts running (give it a few seconds to spin up; wait until the Task status is RUNNING) it will load the existing contents of the topic into the specified Elasticsearch index. In the Elasticsearch console you’ll see

[2017-07-10T13:43:27,164][INFO ][o.e.c.m.MetaDataCreateIndexService] [lGXYRzd] [mysql-foobar] creating index, cause [api], templates [], shards [5]/[1], mappings []  
[2017-07-10T13:43:27,276][INFO ][o.e.c.m.MetaDataMappingService] [lGXYRzd] [mysql-foo/7lXMoHxWT6yIvmxIBlMB9A] create_mapping [type.name=kafka-connect]

Querying the data using the Elasticsearch REST API shows that the data is being streamed to it from Kafka:

$ curl -s "http://localhost:9200/mysql-foobar/_search"|jq '.hits'
{
 "total": 5,
 "max_score": 1,
 "hits": [
 {
 "_index": "mysql-foobar",
 "_type": "type.name=kafka-connect",
 "_id": "mysql-foobar+0+4",
 "_score": 1,
 "_source": {
 "c1": 4,
 "c2": "bar",
 "create_ts": 1501797647000,
 "update_ts": 1501797647000
 }
 },
[...]

The eagle-eyed amongst you will notice that the _id of the document has been set by Kafka Connect to the topic/partition/offset, which gives us exactly-once delivery courtesy of Elasticsearch’s idempotent writes. Now, this is very useful if we just have a stream of events, but in some cases we want to declare our own key—more on this later.

We can also see that mapping has been created, using the schema of the source MySQL table – another great reason for using the Confluent Schema Registry.

$ curl -s "http://localhost:9200/mysql-foobar/_mappings"|jq '.'
{
 "mysql-foobar": {
 "mappings": {
 "type.name=kafka-connect": {
 "properties": {
 "c1": {
 "type": "integer"
 },
 "c2": {
 "type": "text"
 },
 "create_ts": {
 "type": "date"
 },
 "update_ts": {
 "type": "date"
 }

Using Kibana we can easily see the actual data that’s flowing through from Kafka, with the datatypes preserved and thus the timestamps available for selecting and aggregating our data as we want:

Recap

With a few simple REST calls, we’ve built a scalable data pipeline, streaming data from a relational database through to Elasticsearch, and a flat file. With the Kafka Connect ecosystem we could extend and modify that pipeline to land data to HDFS, BigQuery, S3, Couchbase, MongoDB … the list goes on and on!

Stay tuned for more posts in this series that will take a look at some of the additional cool features available to us in Apache Kafka and Confluent Platform.

Other Posts in this Series:

Part 1: The Simplest Useful Kafka Connect Data Pipeline In The World … or Thereabouts – Part 1
Part 3: The Simplest Useful Kafka Connect Data Pipeline in the World…or Thereabouts – Part 3

  • Robin is a Principal DevEx Engineer at Decodable as well as an Oracle Groundbreaker Ambassador. His career has always involved data, from the old worlds of COBOL and DB2, through the worlds of Oracle and Hadoop, and into the current world with Kafka. His particular interests are analytics, systems architecture, performance testing and optimization. He blogs at http://cnfl.io/rmoff and http://rmoff.net/ and can be found tweeting grumpy geek thoughts as @rmoff. Outside of work he enjoys drinking good beer and eating fried breakfasts, although generally not at the same time.

Did you like this blog post? Share it now