Live read models (replicators)

[Martin Krasser]

Hi,

I just created a pull request that adds live read models (as discussed in this thread) to akka-persistence. Live read models can be implemented by extending the Replicator trait (and implementing Actor.receive, see also this example and tests). A replicator is a read-only processor that automatically receives persistent messages that are written by a corresponding processor. It can construct any replicator-specific state from the replicated message stream and even create replicator-specific snapshots. You can think of replicators as the R in CQRS.

Replicators do not directly interact with their corresponding processor, they are updated by running incremental message replays via the journal. Hence, you can start/use replicators also if the corresponding processor is not running. If that processor is started later, asociated replicators are updated automatically. A replicator runs incremental replays periodically based on a configurable interval but applications may also trigger incremental replays at any time.

Replicators can also be used in combination with channels (see example), in the same way as Processor and EventsourcedProcessor.

The previous Processor implementation is now splitted into traits Recovery, Snapshotter and Processor, so that common code can now be shared by Replicator and Processor. The pull request still misses the Java API and user docs. I'll add them later.

Please let me know what you think, would be great to have replicators in 2.3.0.

Cheers,
Martin

-- 
Martin Krasser

blog:    http://krasserm.blogspot.com
code:    http://github.com/krasserm
twitter: http://twitter.com/mrt1nz

[Martin Krasser]

The were two wrong links in the previous post. Here are the correct links to the example and the Snapshotter trait.