-
Notifications
You must be signed in to change notification settings - Fork 3
Batchers
#Batchers The Batchers are responsible of forming the batch of required size from the data coming to batcher.
The batchers are based on time window. For example, if the decided time window is one minute long, every Particle coming to the batcher is checked whether it is the part of the last coming minute. If it is within the minute, it takes its place within the batch based on the timestamp (the ordering is guaranteed after forming the whole batch). If the Particle coming to the Batcher is from the next time window (in our case - from the next minute), first Particle in batch queue is transferred to the next bolt. Therefore, we have sliding time window, within which we order the Particles based on time.
If the Particle comes too late (when the sliding window is ahead on more than one minute), it is completely rejected, i.e., it is not processed and does not follow the topology further.
//TODO: interface code set and javadoc link