Skip to content

framiere/mdc-with-replicator-and-regexrouter

Repository files navigation

MDC and single views

Objective

We have 2 regions: US and EUROPE each have a topic with sales that happened regionaly. We want on each region to have a way to see all sales in all regions.

Replicator

Replicator setup

Why is it that we need to set topic.config.sync=false ?

In this example, if EUROPE_sales and US_sales do not share the same setup... what would be sales topic like ? It's now up to you to sync these, and not replicator. If you want no have this problem, just use Replicator in its default mode :)

Using rename-format

https://docs.confluent.io/current/connect/kafka-connect-replicator/configuration_options.html#destination-topics

This setup is executable in rename-format.sh

Using RegexRouter

https://docs.confluent.io/current/connect/transforms/regexrouter.html

you will need to specify:

  • topic.config.sync=false
  • topic.auto.create=false
  • topic.preserve.partitions=false

This setup is executable in regexrouter.sh

Using the default mode

Simplest replicator setup

This setup is executable in simplest.sh

The main point is that you can subscribe to a list of topics or to a regex.

It's native to the consumer protocol https://docs.confluent.io/current/clients/javadocs/org/apache/kafka/clients/consumer/KafkaConsumer.html#subscribe-java.util.regex.Pattern-

It is also there to be leveraged in the kafka-console-consumer

docker-compose exec broker-europe kafka-console-consumer --bootstrap-server broker-europe:9092 --whitelist "sales_.*" --from-beginning --max-messages 20 --property metadata.max.age.ms 30000

Why specifying metadata.max.age.ms ?

We specify metadata.max.age.ms 30000 (30s) because the default is 5 minutes, we want new topics to be discovered faster.

Why using this method ?

It does simplify everything as there is

  • no duplication
  • simpler reasoning
  • simpler monitoring
  • simpler ACL management
  • less data traveling around
  • less connectors
  • native
  • ...

Using the default mode and having a local merge

Basically, same as the previous solution, plus we merge all the local sales_ topics to a single sales topic.

As a follow of the previous --whitelist "sales_.*" property, the connector will use the topic.regex property. We used metadata.max.age.ms for the client, we will use topic.poll.interval.ms to specify the discovery frequency.

Beware that while it seems simples, there is indeed

  • duplication
  • it kind of breaks the ACLs mindset, while it also can be useful

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Languages