Tuning Process Flow

This section describes the process flows of the wordcount, identity, and fixwindow test cases.

• wordcount

  

  The data is read from a topic of Kafka and is split in the downstream SplitStreamBolt. Only the IP address information in the data is used. The IP address information and the message time are transferred as a pair of values to the WordCountBolt. This bolt collects statistics on received IP address information, calculates the difference between the current system time and the message generation time to obtain the delay of the data stream, and uses the kafkaReporter class to write the result to the topic corresponding to Kafka.

• identity

  

  The data is read from a topic of Kafka and sent to the downstream IdentityBolt. This bolt only parses the received messages and does not perform other processing. Then, the data stream delay is obtained in the same way as the wordcount case and the data is sent to the corresponding topic.

• fixwindow

  

  The data is read from a topic of Kafka and sent to the downstream ParserBolt. This bolt parses the received messages and sends it to the SlidingWindowBolt. This bolt processes all data in a window of a fixed size at a time and performs the count operation on the data. The delay of each operation is the system time when the operation is complete minus the earliest generation time of all data in the window. The kafkaReporter class is used to write the result to the topic corresponding to Kafka.