Towards Better Engineering of Enterprise Resource Planning Systems
In spite of their high implementation failure rate, Enterprise Resource Planning (ERP) software remains a popular choice for most businesses. When it succeeds, ERP software provides effective integration of formerly isolated multiple systems. This integration yields significant business efficiencies...
| Main Author: | |
|---|---|
| Format: | Others |
| Published: |
North Dakota State University
2016
|
| Online Access: | http://hdl.handle.net/10365/25580 |
| Summary: | In spite of their high implementation failure rate, Enterprise Resource Planning (ERP) software remains a popular choice for most businesses. When it succeeds, ERP software provides effective integration of formerly isolated multiple systems. This integration yields significant business efficiencies. Replacing legacy systems with ERP software requires a great many trade-offs. We found that using bipartite graph can facilitate requirements elicitation in ERP procurement. It can have great impact on different activities in subsequent phases. Such activities include product line engineering, domain analysis, test-driven development and knowledge reuse in system development life cycle (SDLC). Bigraph representation of legacy and ERP requirements also helps determining the enhancement needs. It provides a better control for stakeholders to decide the scope for development and testing. ERP streamlines operations and flow of information across an organization. Business functions in ERP system are always master-data driven. Migration of data from legacy to ERP is therefore a critical success factor for ERP procurement projects. Correct data conversion is essential for integration and acceptance testing. It is a complex procedure and requires additional effort because of large volume of data. The architecture and design of data conversion process must ensure the referential integrity among different business modules. Our process for data conversion starts with a test-first approach. Next it conducts execution of conversion programs in parallel. Our parallel approach replaces the old sequential style of execution. Through this process we could correct data mapping errors and data anomalies before conversion. Parallelized execution drastically reduces the time needed for execution of conversion programs and provides more time for testing. We verified the feasibility of our approach by multiple industrial projects. === National Institutes of Health (NIH) |
|---|