Hi,
I just pushed a prototype [1] for integrating Oak observation with
Apache Kafka [2] to my private GitHub account. This is a prove of
concept implementation of an Oak Observer sending all changes in Oak
repository to a Kafka message log. It should be possible to implement
JCR's ObservationManager interface on top of a Kafka message log on the
consumer side. But this is currently left as an exercise ;-)
The prototype comes with an OSGi bundle, which can be directly deployed
into a running Sling instance. See the README [3] for further details.
Kafka could help us in cases where
- messages should be persisted,
- processing should take place out of band (e.g because it is too
expensive),
- scaling is important as Kafka supports partitioning the message log
across multiple servers,
- availability is important as Kafka supports failing over to secondariness.
Michael
[1] https://github.com/mduerig/oak-kafka
[2] http://kafka.apache.org/
[3] https://github.com/mduerig/oak-kafka/blob/master/README.md
