Score: 1

Object Abstraction To Streamline Edge-Cloud-Native Application Development

Published: December 27, 2025 | arXiv ID: 2512.22534v1

By: Pawissanutt Lertpongrujikorn

Potential Business Impact:

Simplifies building apps in the cloud.

Business Areas:
PaaS Software

Cloud computing has fundamentally transformed application development, yet a gap remains between the serverless promise of simplified deployment and its practical realization due to fragmentation across function runtimes, state management, and orchestration. This dissertation addresses this gap through empirical validation and technical innovation, establishing the Object-as-a-Service (OaaS) paradigm as a unified approach to cloud-native development. Grounded in evidence from three studies - practitioner interviews (21 participants), a human study on developer experience (39 participants), and NSF I-Corps customer discovery (101 interviews across 86 organizations) - this work demonstrates that infrastructure complexity taxes productivity, with practitioners prioritizing automation and maintainability over cost optimization. The dissertation makes five major contributions: (1) the OaaS paradigm unifies resource, state, and workflow management via the Oparaca prototype, demonstrating negligible overhead and state-of-the-art scalability; (2) SLA-driven OaaS enables declarative management of non-functional requirements like availability, consistency, and latency; (3) OaaS-IoT with EdgeWeaver extends the paradigm to the edge-cloud continuum, achieving 31% faster task completion and a 44.5% reduction in lines of code compared to traditional FaaS; (4) commercialization validation establishes a pathway targeting technology SMEs and startups; and (5) an empirical methodology for grounding technical research in validated practitioner needs. By consolidating fragmented abstractions and automating performance optimization, OaaS establishes a foundation for cloud-native platforms that hide infrastructure complexity and empower developers to focus on innovation.

Repos / Data Links

Page Count
181 pages

Category
Computer Science:
Distributed, Parallel, and Cluster Computing