ESR 12: Heterogeneous Low-Code Model Query Optimisation
Your TeamYou will be hired by University of York (United-Kingdom).
As software systems become more complex, underlying models in LCEPs grow proportionally in both size and complexity. Such models can be persisted in a variety of proprietary or standard formats (such as XMI), and in different types of back-ends (e.g. file systems, relational databases, document databases). High-level, concise and tailored model query languages such as OCL and EOL can be used to shield query developers from the intricacies of the underlying model formats/back-ends but this typically has a significant impact on performance. Recently, we have shown how sophisticated runtime query optimisation can be used to drastically improve the execution time of high-level OCL-style queries executed over models stored in relational and non-relational databases. The objectives of this project are to: (1) investigate the applicability of runtime query optimisation techniques to a wide range of model persistence formats and back-ends, (2) identify reusable optimisation primitives and patterns across different formats and back-ends, and (3) evaluate the obtained benefits in terms of performance and memory footprint.
This project will produce novel techniques and algorithms for optimisation of queries operating on low-code systemmodels captured using differentmodelling languages andmodel representation formats. It will also produce an open-source prototype that will implement the identified algorithms and techniques on top of existing model query languages. While the precise performance benefits will depend on the nature of individual queries and the underlying model representation formats, based on our preliminary results in we expect an increase of at least one order of magnitude in query execution time for certain classes of queries (e.g. filtering all instances of a type).
This position has been filled.
Please note that the vacancy on the institutionnal website must be considered as the official version of this PhD position.