Improved SOA persistence architectural model

Companies often face difficulties when building the data services layer. The most common problems are poor data quality, perfor-mance issues and heterogeneity of data sources. In this article, we propose solutions to the problems described and merge them into a comprehensive architectural model for the storage and manage-ment of permanent data in SOA (Service Oriented Architecture). In order to overcome problems related to poor data quality, our archi-tectural model requires the use of adjusted MDM (Master Data Management). Furthermore, we propose that all data transfer be-tween data services and business level goes in form of SDO (Ser-vice Data Objects), which allow us a uniform data access and manipulation, regardless of the data type. As the best approach to improve performances, scalability, availability and reliability of the data services layer, our architecture introduces caching in the in-termediate layer with the use of in-memory data grids. As an ex-ample of how the proposed solution can also be used to improve parts of the SOA infrastructure, we will present possible approach to improve BPEL (Business Process Execution Language) dehy-dration store.