Banca de DEFESA: JÚLIO GUSTAVO SOARES FIRMO DA COSTA

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : JÚLIO GUSTAVO SOARES FIRMO DA COSTA
DATE: 03/06/2026
TIME: 09:00
LOCAL: on-line
TITLE:

Multi-tenant computing service for data persistence on low-code platforms

KEY WORDS:

Runtime Model Interpretation; Low-Code Platforms; Multi-Tenancy; Model-Driven Engineering; Cloud Computing; Resource Efficiency

PAGES: 195
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Sistemas de Informação
SUMMARY:

Cloud computing operates under a structural paradox: despite more than two decades of technological maturity, its services still exhibit utilization rates between 30% and 50%, with annual waste estimated at US$ 17 billion and projected consumption of 10% of the global electricity supply by 2030. In the field of Software Engineering, low-code platforms have emerged as a response to the growing demand for rapid system construction, particularly in small and medium-sized enterprises. However, when implementing advanced architectural patterns, these platforms generate a dedicated service instance for each client, perpetuating at the operational level the very resource waste they were meant to address.

This thesis investigates runtime model interpretation as an architectural alternative to that paradigm. The central hypothesis is that generic interpretation engines, processing declarative domain and process models, can materialize advanced patterns without specialized code generation and, thereby, enable Multi-Tenant Single-Instance (MTSI) operation --- a mode in which a single service instance serves multiple clients simultaneously. To investigate this hypothesis, the research empirically characterized the costs of conventional implementations in a realistic domain of vehicular incident support, proposed an interpretation platform organized into specialized engines, and compared the two paradigms under equivalent load regimes in a Kubernetes environment.

The results validate the operational viability of runtime model interpretation and evidence the perceived potential of the architecture for Multi-Tenant Single-Instance (MTSI) operation. The interpretation platform consumed resources modestly higher than those required by a single conventional instance dedicated to a single client, with a 100% success rate across more than seven thousand six hundred requests and absolute latency within margins compatible with interactive applications (median below 10 ms, p95 below 60 ms). Anchored in real operational data --- a typical portfolio of approximately 6500 insured vehicles generating 22 to 27 incident notifications per month ---, the architecture points to potential reduction in per-tenant resource consumption compared to the prevailing paradigm based on dedicated instances (MTMI), under which each new client requires an independent cluster. The empirical characterization also documented that conventional implementations multiply artifacts by factors ranging from 19 to 42 times the underlying domain concepts, with 40% to 48% contamination between domain and infrastructure layers---quantifying a technical debt historically treated as an abstract concern.

This thesis contributes on three complementary fronts. It presents an architectural framework, grounded on six principles, for building interpretation-based platforms. It provides auditable metrics for the technical debt embedded in advanced architectural patterns, enabling rational cost-benefit analyses. And it demonstrates, under controlled conditions, that runtime model interpretation can reconcile development agility, operational efficiency, and socio-environmental responsibility in the contemporary context of cloud computing.

COMMITTEE MEMBERS:
Presidente - 1673543 - SAMUEL XAVIER DE SOUZA
Interno - 1345674 - LUIZ MARCOS GARCIA GONCALVES
Externo ao Programa - 2453033 - ANDERSON LUIZ DE OLIVEIRA CAVALCANTI - UFRNExterno ao Programa - 1678918 - NELIO ALESSANDRO AZEVEDO CACHO - UFRNExterna à Instituição - ELISA YUMI NAKAGAWA - USP
Externo à Instituição - FERNANDO ANTONIO MOTA TRINTA - UFC