Difference between Apache Beam and Apache Nifi

问题:

What are the use cases for Apache Beam and Apache Nifi? It seems both of them are data flow engines. In case both have similar use case, which of the two is better?

回答1:

Apache Beam is an abstraction layer for stream processing systems like Apache Flink, Apache Spark (streaming), Apache Apex, and Apache Storm. It lets you write your code against a standard API, and then execute the code using any of the underlying platforms. So theoretically, if you wrote your code against the Beam API, that code could run on Flink or Spark Streaming without any code changes.

Apache NiFi is a data flow tool that is focused on moving data between systems, all the way from very small edge devices with the use of MiNiFi, back to the larger data centers with NiFi. NiFi's focus is on capabilities like visual command and control, filtering of data, enrichment of data, data provenance, and security, just to name a few. With NiFi, you aren't writing code and deploying it as a job, you are building a living data flow through the UI that is taking effect with each action.

Stream processing platforms are often focused on computations involving joins of streams and windowing operations. Where as a data flow tool is often complimentary and used to manage the flow of data from the sources to the processing platforms.

There are actually several integration points between NiFi and stream processing systems... there are components for Flink, Spark, Storm, and Apex that can pull data from NiFi, or push data back to NiFi. Another common pattern would be to use MiNiFi + NiFi to get data into Apache Kafka, and then have the stream processing systems consume from Kafka.