December 2017
Beginner to intermediate
322 pages
6h 43m
English
Building redundant, fault-tolerant, and highly performant stream processing is hard. In traditional batch processing, when a job fails, we just retry or re-run either the failed components or a small portion of the failed job. The batch job world operates on the assumption that data is fixed and finite. The underlying infrastructure is built on that immutable data model. The worst case scenario is that the batch job has to be run from the beginning to the end again.
Similar to traditional business analytics solutions, the architecture for real-time event processing is based on the pattern of data ingestion, processing, and ultimately notification/presentation by either the end users or other applications. ...
Read now
Unlock full access