Some Hadoop components you only need to have it on one server. In a typical enterprise with a production Hadoop ecosystem, you might have 10 servers with HBase, five with storm etc. For this list, which of the following would be installed on many servers and which would be on just one or two? I would expect Kafka to be on many servers. I know HDFS would be on many servers, that's why it is not on the list.
Many servers or just a few?