Basically, the Hadoop initially was designed for batch, not real time processing. However, the answer to your questions is heavily depended on what exactly you are doing to do with Hadoop. In case you would like to work with real time, real time events, streams etc. I strongly suggest you to look at the Spark (https://spark.apache.org/), Strom (https://storm.apache.org/) and may be Kafka (http://kafka.apache.org/) or Flume (https://flume.apache.org/).