Partitioner is used to distribute events over partitions.
Currently, there are two implemented partitioners:
- RandomPartitioner uses round robin algorithm,
- HashPartitioner uses hash function
Special ShardingKey is used to determine which tags of the Event should be used in hash function. Note, it's possible to use nested Containers. Thus, ShardingKey consists of a sequence of tag paths in HPath form. See example below:
topLevelTag/secondLevelTag/stringTag
Here, path consists of two nested containers (topLevelTag and secondLevelTag) and leaf tag stringTag.
Those paths are defined in Stream or Timeline entities.
Hash function is defined by Hasher's implementation. See NaiveHasher - the naive hash function implementation.