Grant: Alliance Grants
University of Ontario Institute of Technology
$27,450.00
Developer Recommender for Cloud-Native Applications for Fault Analysis & Localization
FY 2023-24
Summary
This project investigates Large Language Models (LLM) as recommendations for run time faults in cloud-native microservice applications and their integration into a software development cycle. These run time faults typically expose themselves as performance degradation faults in the application and, as such, do not surface as software failures that are simpler to detect using code analysis. In prior work on this project, we have developed tools that facilitate the detection and identification of lock contention faults based on the definition of lock contention anti-patterns. These lock contention anti-patterns are associated with a set of recommendations and code refactoring suggestions. Recently, the work in LLMs has shown some promise as suitable recommendation tools as their natural ability to understand software code and respond to user prompts is rather impressive though the validity of the results is highly dependent on how well the LLM is tuned for particular faults as well as the style of prompts. By enhancing software fault detection in complex, multi-threaded applications, this project aims to significantly reduce the development time of microservice based software applications that are leveraged in digital services such as healthcare, banking, and e-commerce in Canada. These improvements in reliability and efficiency can lead to increased trust and adoption of digital technologies, enabling economic growth. Our focus on training LLMs for software faults, particularly for lock contention, addresses a crucial gap in the lack of recommendation tools for software runtime faults, setting the stage for more resilient and efficient software solutions. As a recommendation system, we envision that these algorithms could be applied in two modes: interactively in an Integrated Development Environment (IDE) as part of the developer's coding workflow or as a standalone tool that produces a report of potential issues and suggested fixes. They may be utilized as part of run times and tools for microservice frameworks or be integrated into an AI-based platform to manage the Software Development Life Cycle (SDLC.)
Program
Alliance Grants
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
590675-2023
Name nom
Liscano, Ramiro R
Department département
Electrical, Computer and Software Engineering
Organization id
19222
Institution établissement
University of Ontario Institute of Technology
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2023
Award amount
27450
Program id
ALLRP
Program name en
Alliance Grants
Program name fr
Subventions Alliance
Group en
RESEARCH PARTNERSHIPS
Group fr
PARTENARIATS DE RECHERCHE
Committee code
96
Committee name en
RPP Internal Decision Cttee
Committee name fr
Cté de décision interne - PPR
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2705
Research subject group en
Information technology
Research subject group fr
Software and development
Research subject en
Technologie de l'information
Research subject fr
Logiciels et développement
Application title
Developer Recommender for Cloud-Native Applications for Fault Analysis & Localization
Keyword
cloud-native micro-service; large language models; software fault recommendation
Application summary
This project investigates Large Language Models (LLM) as recommendations for run time faults in cloud-native microservice applications and their integration into a software development cycle. These run time faults typically expose themselves as performance degradation faults in the application and, as such, do not surface as software failures that are simpler to detect using code analysis. In prior work on this project, we have developed tools that facilitate the detection and identification of lock contention faults based on the definition of lock contention anti-patterns. These lock contention anti-patterns are associated with a set of recommendations and code refactoring suggestions. Recently, the work in LLMs has shown some promise as suitable recommendation tools as their natural ability to understand software code and respond to user prompts is rather impressive though the validity of the results is highly dependent on how well the LLM is tuned for particular faults as well as the style of prompts. By enhancing software fault detection in complex, multi-threaded applications, this project aims to significantly reduce the development time of microservice based software applications that are leveraged in digital services such as healthcare, banking, and e-commerce in Canada. These improvements in reliability and efficiency can lead to increased trust and adoption of digital technologies, enabling economic growth. Our focus on training LLMs for software faults, particularly for lock contention, addresses a crucial gap in the lack of recommendation tools for software runtime faults, setting the stage for more resilient and efficient software solutions. As a recommendation system, we envision that these algorithms could be applied in two modes: interactively in an Integrated Development Environment (IDE) as part of the developer's coding workflow or as a standalone tool that produces a report of potential issues and suggested fixes. They may be utilized as part of run times and tools for microservice frameworks or be integrated into an AI-based platform to manage the Software Development Life Cycle (SDLC.)
