Jan Bednar created CAMEL-15640:
----------------------------------
Summary: Partition component/EIP
Key: CAMEL-15640
URL: https://issues.apache.org/jira/browse/CAMEL-15640
Project: Camel
Issue Type: New Feature
Reporter: Jan Bednar
Assignee: Jan Bednar
Sometimes you need to process messages in parallel and also ensure order of
messages per key. Kafka is great for this with its partitioning and ActiveMQ
with message groups, but in small single instance project may be broker
overkill.
I needed that feature in one project and solved it with something what doesnt
look nice, similar to this pseudo code:
{code:java}
final int consumers = 10;
from("direct:source")
.process(exchange ->
exchange.getMessage().setHeader("partition",
exchange.getMessage().getHeader("partition_key")
.hashCode() % consumers)
)
.toD("seda:partition_${header.partition}");
for (int i = 0; i < consumers; i++) {
from("seda:partition_" + i)
.to("direct:target");
}
{code}
It would be great to have something like Partition component (*or maybe better
Partition EIP?*)
It can be similar to seda, but with following features.
* Pluggable PartitionStrategy with pluggable Expression for partition key.
Default can be HashPartitionStrategy.
* Message will be assigned to Thread based on following algorithm:
_target_thread_id = hash % concurrentConsumers_
*Community feedback/discussion about this feature welcomed.*
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
