I have a single master topic and multiple predicates each of which has an output topic associated with it. I want to send each record to ALL topics that whose predicate resolves to true. I am using Luwak to test which predicates a record satisfies (to use this library you evaluate a document with a list of predicates and it tells you which ones matched - i.e. I only call it once to get the list of satisfied predicates).
I am trying to use Kafka Streams for this but there doesn't seem to be the appropriate method on KStream (KStream#branch only routes a record to a single topic).
One possible approach is as follows:
Stream from master
Map the values into a format with the original content and the list of matching predicates
Stream to an intermediate with-matches topic
For each predicate/output topic
Stream from intermediate with-matches topic
Filter "does list of matches predicates contain predicate ID"
Map the values to just the original content
Stream to corresponding output topic
Such an intermediate topic seems "clunky" though. Any better suggestions?
I am using:
Kafka is able to seamlessly handle multiple producers that are using many topics or the same topic. The consumer subscribes to one or more topics and reads the messages. The consumer keeps track of which messages it has already consumed by keeping track of the offset of messages.
A. Yes, Kafka's design allows consumers from one consumer group to consume messages from multiple topics.
Multi-Topic ConsumersWe may have a consumer group that listens to multiple topics. If they have the same key-partitioning scheme and number of partitions across two topics, we can join data across the two topics.
You can simple apply multiple filters in parallel to the same KStream
instance:
KStream stream = ...
stream.filter(new MyPredicate1()).to("output-topic-1");
stream.filter(new MyPredicate2()).to("output-topic-2");
stream.filter(new MyPredicate3()).to("output-topic-3");
// ... as as many as you need
Each record will be sent to each predicate once -- it's conceptually a broadcast to all filters, but records will not be physically replicated, so there is no memory overhead.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With