In this architectural style, the system is unidirectional. One module follows the other. Pipes connect these modules, which can be a producer (feeds the pipeline with data), filter (manipulates the data that it receives), or consumer (which is the end of the pipeline and can persist the data in a database).
The pipeline architecture is a common strategy in Unix development using pipes |.
The pipeline architecture is simple and cheap since it’s monolithic by nature and doesn’t have distributed system complexity.
It’s fairly testable because of its partial modularity.
The worst part is it’s not fault-tolerant. One of the failures in any part causes a complete system failure. Also, it’s not scalable and elastic. Although it’s possible to make certain functions scale more than others in the monolithic architecture, this operation is very complex.
Preview: