Batch Terminology

The following table briefly describes the terminology used in batch.

Term	Description
Batch Job (JOB)	A unit of execution and re-execution of a batch job composed of a flow to process batch business, which is written as an XML configuration that describes the flow. Commonly, large-scale batch jobs consist of complex processing procedures including calls to external programs, but batch jobs defined in the batch application are limited to flows that call unit Steps implemented by the batch application.
Step	The minimum unit of business implementation that constitutes a batch job. It operates only as a component of a batch job and cannot be executed independently. When a batch job is re-executed, individual Steps may be re-executed or excluded from execution depending on conditions.
Item	The minimum unit of data to be processed in a batch Step according to business requirements. It is extracted per record from DB or file resources. It is generally converted and delivered as an IO-type data object, and especially in file processing, it must be used in IO form in order to use the message conversion function of IO.
ItemReader	Performs the role of acquiring data to be processed in a batch Step from resources such as DB or files and delivering it to ItemProcessor or ItemWriter. When fetching items from the DB, the business developer directly writes the ItemReader using DBIO. In contrast, when fetching items from file resources, the file reader XML provided by the batch framework is used. However, if the file reader XML cannot be used, the business developer must implement it directly.
ItemProcessor	Batch business developers need to clearly separate the role into the procedure that processes/transforms business data according to business requirements and the procedure that records or transmits the resulting data, and in this case they can define a separate ItemProcessor to perform the processing.
ItemWriter	Performs roles such as recording items delivered from ItemReader or ItemProcessor to storage media such as DB or files, or transmitting batch processing result data per item to external systems. It may also include the role of a processor that processes items according to business requirements. To record items in the DB, the business developer directly writes the ItemWriter using DBIO, and to record items to file resources, the file writer XML provided by the batch framework is used. However, if the file writer XML cannot be used, the business developer must implement it directly.
Scope	Refers to the period from when an object is created until it is destroyed. The creation and destruction of the above-mentioned ItemReader, ItemProcessor, and ItemWriter objects are handled by the actual framework, so for business developers it can be understood as the period during which the instance of the relevant object remains available. The batch framework provides a scope called "step", which can be used only in the batch application, and this means that the object is maintained and its member data is preserved for the period during which a single Step is executed.
Item-Oriented Processing	The processing procedure of a batch Step consists of extracting items using ItemReader and then sequentially passing them to ItemProcessor and ItemWriter according to the Step definition for processing and recording. In this way, Item-Oriented Processing refers to the procedure in a framework-based batch Step that uses items as the minimum processing unit.
Chunk-Oriented Processing	The act of finally persisting data processing and recording to the medium; that is, writing to a file (write-flush) or committing a data transaction is performed by intervals in consideration of batch processing performance. At this time, the unit of data persistence is defined by a Step attribute called "commit-interval". As seen in "Item-Oriented Processing", data processing is implemented on an item basis, while persistence is performed by interval, so a mismatch occurs between the processing unit and the persistence unit, which requires additional implementation for adjustment. To address this, the framework implements a strategy in which items extracted by ItemReader and processed by ItemProcessor are delivered to ItemWriter as persistence units using an internal buffer. In this process, the framework can perform the implementation of adjustment work due to persistence failures, and this processing type is defined as "Chunk-Oriented Processing".

Term

Description

Batch Job (JOB)

A unit of execution and re-execution of a batch job composed of a flow to process batch business, which is written as an XML configuration that describes the flow. Commonly, large-scale batch jobs consist of complex processing procedures including calls to external programs, but batch jobs defined in the batch application are limited to flows that call unit Steps implemented by the batch application.

Step

The minimum unit of business implementation that constitutes a batch job. It operates only as a component of a batch job and cannot be executed independently. When a batch job is re-executed, individual Steps may be re-executed or excluded from execution depending on conditions.

Item

The minimum unit of data to be processed in a batch Step according to business requirements. It is extracted per record from DB or file resources. It is generally converted and delivered as an IO-type data object, and especially in file processing, it must be used in IO form in order to use the message conversion function of IO.

ItemReader

Performs the role of acquiring data to be processed in a batch Step from resources such as DB or files and delivering it to ItemProcessor or ItemWriter. When fetching items from the DB, the business developer directly writes the ItemReader using DBIO. In contrast, when fetching items from file resources, the file reader XML provided by the batch framework is used. However, if the file reader XML cannot be used, the business developer must implement it directly.

ItemProcessor

Batch business developers need to clearly separate the role into the procedure that processes/transforms business data according to business requirements and the procedure that records or transmits the resulting data, and in this case they can define a separate ItemProcessor to perform the processing.

ItemWriter

Performs roles such as recording items delivered from ItemReader or ItemProcessor to storage media such as DB or files, or transmitting batch processing result data per item to external systems. It may also include the role of a processor that processes items according to business requirements. To record items in the DB, the business developer directly writes the ItemWriter using DBIO, and to record items to file resources, the file writer XML provided by the batch framework is used. However, if the file writer XML cannot be used, the business developer must implement it directly.

Scope

Refers to the period from when an object is created until it is destroyed. The creation and destruction of the above-mentioned ItemReader, ItemProcessor, and ItemWriter objects are handled by the actual framework, so for business developers it can be understood as the period during which the instance of the relevant object remains available. The batch framework provides a scope called "step", which can be used only in the batch application, and this means that the object is maintained and its member data is preserved for the period during which a single Step is executed.

Item-Oriented Processing

The processing procedure of a batch Step consists of extracting items using ItemReader and then sequentially passing them to ItemProcessor and ItemWriter according to the Step definition for processing and recording. In this way, Item-Oriented Processing refers to the procedure in a framework-based batch Step that uses items as the minimum processing unit.

Chunk-Oriented Processing

The act of finally persisting data processing and recording to the medium; that is, writing to a file (write-flush) or committing a data transaction is performed by intervals in consideration of batch processing performance. At this time, the unit of data persistence is defined by a Step attribute called "commit-interval". As seen in "Item-Oriented Processing", data processing is implemented on an item basis, while persistence is performed by interval, so a mismatch occurs between the processing unit and the persistence unit, which requires additional implementation for adjustment. To address this, the framework implements a strategy in which items extracted by ItemReader and processed by ItemProcessor are delivered to ItemWriter as persistence units using an internal buffer. In this process, the framework can perform the implementation of adjustment work due to persistence failures, and this processing type is defined as "Chunk-Oriented Processing".