Hi! Our architecture real time processing based on storm. I know Hadoop is prefered to aggregation, but... 1. In our architecture we put the source data to ElasticSearch because we want to search in the source data no only the aggregated 2. In this case we not using the Hadoop HDFS (why store same data in two different location), so why we need to use Hadoop to reduce/map operation instead of a second storm topology? 3. (The aggregated data written back to ES so we not store this in HDFS)
In our case Hadoop still prefered to aggregation? Or replaceable with a storm topology? Thanks b0c1
