Summary
As this chapter approaches its end, you must have got a brief idea about the Nimbus, supervisor, UI, and Zookeeper processes. This chapter also taught you how to tune parallelism in Storm by playing with the number of workers, executors, and tasks. You became familiar with the important problem of distributing computation, that is, failures and overcoming failures by different kinds of fault tolerance available in the system. And most importantly, you learned how to write a "reliable" spout to achieve guaranteed message processing and linking in bolts.
The next chapter will give you information about how to build a simple topology using a Python library called Petrel. Petrel addresses some limitations of Storm's built-in Python support, providing simpler and more streamlined development.
