Archiving with Athamas: A Framework for Optimized Handling of Domain Knowledge

摘要

Continuous changes in requirements of a multi-tiered project can significantly deteriorate the maintenance process. This is mainly due to incompatibility between application design and newly introduced requirements. This paper presents a potential application framework, Athamas, which provides a scalable way to agilely adapt to changing data requirements. A currently functional initiative of Athamas is to allow domain knowledge generation from application components without users or business processes dictating data handling and integration requirements. This development allows the framework to be used 1) today as a use case alternative to relational databases for archiving domain knowledge into storage containers and 2) in the future for optimally extracting the knowledge from the storage containers. Evaluations of Athamas-based applications are made against applications using MySQL’s MyISAM and ARCHIVE database storage engines for data archival purposes. Athamas using a zlib compression layer significantly reduces the storage size utilization to 3% of MyISAM and 24% of ARCHIVE, along with publishing time improvements by a factor of 4.9 and 2.6 respectively. Using Athamas with bzip2 compression, storage usage is down to 8% of ARCHIVE usage but at a cost of slower performance up to 4.5 times. The paper also discusses future enhancements to Athamas allowing users to efficiently query domain knowledge.