Evolução e tendências em plataformas de processamento de grafos em larga escala: uma revisão sistemática

Abstract

We present a systematic literature review of large-scale graph processing platforms spanning 2013–2023, examining how solutions have evolved to address efficiency, scalability, security, and interoperability. Thirteen high-impact primary studies were selected and classified into four thematic “waves”: (1) consolidation of vertex-centric models with CPU+GPU extensions; (2) I/O optimizations and novel partitioning architectures; (3) fine-grained observability and cost-aware elasticity; and (4) energy sustainability, temporal property graphs, and serverless orchestration. Our synthesis reveals performance gains up to 120× in execution time, 1.9× improvements in TEPS/W, 14× speed-ups via SSD prefetch, and 34 % cost reductions, alongside persistent communication bottlenecks, a near-complete absence of integrated security mechanisms, and a lack of unified benchmarking frameworks. We propose a future research roadmap including distributed graph threat modeling, adaptive streaming partitioning, cross-platform APIs, federated FaaS prototypes, and real-time graph digital twins. This work provides both theoretical insights and practical guidelines for designing next-generation graph processing platforms that balance high performance, resilience, and trustworthiness in heterogeneous cloud-edge environments.