"As Kafka’s popularity grows, enterprises often find themselves deploying not a single cluster, but several clusters. This is usually because of scale, but can be driven by other needs such as activities in different geographies or compliance requirements. Whenever you have multiple clusters, the ability to move data between clusters is very valuable. It enables very different scenarios such as geo-replication, disaster recovery, hybrid cloud architectures, but also simply allows reusing data between applications.
MirrorMaker is the Apache Kafka project’s solution for mirroring clusters and consists of 3 connectors for Connect. This tool allows creating all sorts of mirroring topologies by combining the connectors in different ways. In this session we will demonstrate how to use MirrorMaker by exploring two of the most common mirroring use cases: geo-replication and disaster recovery. For each we’ll explain the combination of connectors to use and highlight the key decision points and configurations that you should carefully consider. At the end of the session you will understand the capabilities of MirrorMaker and the process of building powerful mirroring scenarios with this tool."
29. 29
Which connectors do you need?
What features are required?
What monitoring do you need?
Building a Scenario
30. 30
Which connectors do you need?
What features are required?
What monitoring do you need?
What processes do you need to put in place?
Building a Scenario
51. 51
MirrorMaker Exactly Once - New in 3.5
• exactly.once.source.support on Workers
https://kafka.apache.org/documentation/#connect_exactlyoncesource
53. 53
Source Cluster Tag - New in 3.5
• add.source.alias.to.metrics (default from 4.0)
kafka.connect.mirror:type=MirrorSourceConnector
source=FR
target=UK
topic=FR.sales
partition=0
54. 54
11 Chapters available in early access
https://www.oreilly.com/library/view/kafka-connect/9781098126520/
https://www.redhat.com/en/engage/build-data-pipelines-20220603