TagsDesign patterns are a formalization of common structures and interactions that arise repeatedly from the process of designing a range of applications. The approach of identifying design patterns and documenting them in a standard format is a way to simplify the design process. An understanding of these common design patterns should enable you to use them as building blocks for your particular design. The design of an application can then be constructed as a combination of design patterns rather than a combination of individual Operators. This is easier to understand and means that you are able to benefit from the distilled design wisdom of previous successful applications.
As implementations of Streams grow, you should expect the number and types of design patterns to also grow. The following design patterns have been identified at this time:
- Filter pattern for data reduction
- Outlier pattern for data classification and outlier detection
- Parallel pattern for high volume data transformation
- Pipeline pattern for high volume data transformation
- Alerting pattern for real-time decision making and alerting
- Enrichment pattern for supplementing data
- Unstructured Data pattern for supporting unstructured data analysis
- Consolidation pattern for combining multiple sources
- Merge Pattern for combining similar sources
- Integration Pattern for leveraging existing analytics
###References
[1] Chuck Ballard et. al, IBM InfoSphere Streams: Harnessing Data in Motion, IBM Redbooks, pp.88, September 2010.