Software architecture and design (IV) -- data flow architecture

Data flow architecture

In data flow architecture , The whole software system is regarded as a continuous set or transformation of a set of input data , Data and operations are independent of each other . In this way , Data access system , Then flow through the modules one at a time , Until they are assigned to a final destination （ Output or data storage ）.

The connection between components or modules can be realized as I / O flow ,I / O Buffer , Pipe type or other type of connection . Data can flow in a circular graph topology in a acyclic linear structure or tree structure .

The main goal of this approach is to achieve the quality of reuse and modifiability . It applies to inputs and outputs that involve ordered definitions （ Such as compilers and business data processing applications ） A well-defined series of independent data conversion or computing applications . There are three types of execution sequences between modules -

• Batch order
• Pipeline and filter or non sequential pipeline mode
• process control

Batch order

Batch order is a classical data processing model , Among them, the data conversion subsystem can start its process only after its previous subsystem has completely passed ,

• The overall data that flows from one subsystem to another .

• Communication between modules is carried out through temporary intermediate files that can be deleted by continuous subsystems .

• It is suitable for batch data applications , Each subsystem reads the relevant input file and writes it to the output file .

• Typical applications of this architecture include business data processing , Such as bank and utility billing .
  

advantage
Usually , Batch order provides a simpler division on the subsystem . Each subsystem can be an independent program , Used to process input data and generate output data .

shortcoming