bmap4j - Batch Management And Processing For Java |
Batch Transaction Processing Fundamentals - BTP architecture |
de | |
BTP ArchitekturProcessing of large data amounts in batch-mode (Batch Transaction Processing, BTP) fundamentally differs in various aspects from online-processing (Online Transaction Processing, OLTP). The platforms that IT needs to provide for both of those processing modes need to be adapted depending on the different requirements and therefore also require differentiated architecture. However, common features of both architectures have to be found and combined in a way that results in the greatest possible synergies. BTP ArchitekturThe following graphic demonstrates this using the classical, layer-based OLTP enterprise applications architecture with the following layers: Client, Presentation, BusinessLogic, Persistence, Data.
The two top layers of this existing architecture are now supplemented with the two BTP specific Processing and Management Layers to guarantee the additional functions required for the processing of mass data. The existing BusinessLogic, Persistence and Data Layers remain. The BusinessLogic Layer will potentially require additional features for mass selection and -update. The Processing Layer implements the batch program's technical functionality. For this purpose it accesses the BusinessLogic layer. The Management Layer ensures the batch program's non-functional aspects and supports the Processing Layer. Batch Processing FrameworkFrom our point of view the following factors are crucial to build and operate a BTP platform:
Requirements for a modern BTP PlatformEssential requirements for a BTP platform are therefore as follows: Needless to say, basic services like safety, scalability and concurrency have to be assured. However, the requirements are not fundamentally different compared to those of an OLTP systems and should therefore ideally be provided for by an existing enterprise application platform. |
||
25.05.2018 12:24 | Copyright © 2007 - 2018 by Xinventa GmbH, Luzern |