Explanation:
The following two applications are best suited to batch processing:
- Employee payroll processing and generating payroll checks
- Setting inventory stocking levels based on seasonal sales volume
Data for batch processing is collected over time, often from different data sources, and it is processed as a dataset that includes a range of rows or all rows in the dataset. Batch processing is designed to handle processing of large datasets. There is typically a long latency between data collection and data processing.
The following two applications are best suited to stream processing:
- Reporting the number of users and bandwidth user for an online game
- Identifying detected manufacturing errors to automatically reject failing parts Stream processing is designed for real-time or near real-time data processing, often as a data load process with minimal processing. Data must be able to stream out as quickly as it streams in for processing. Data is either processed as it is generated or in micro-batches of a very few rows with latency of no more than a few milliseconds.