Sarcouncil Journal of Multidisciplinary

Abstract: Serverless computing architectures have fundamentally transformed large-scale data processing workflows, introducing novel orchestration challenges that require sophisticated parallelization strategies. This article evaluates elastic parallelism strategies in serverless orchestration using AWS Step Functions through controlled experimental evaluation of three distinct approaches: static batching, dynamic subtree partitioning, and adaptive Map-State chunking. The experimental framework employs a multi-terabyte synthetic dataset emulating legal contract processing workflows characterized by heterogeneous document characteristics and variable transformation requirements. Performance evaluation reveals that adaptive chunking strategies demonstrate superior execution characteristics compared to traditional static approaches, achieving substantial latency reductions and cost optimizations across diverse workload conditions. The implementation incorporates AWS Step Functions' native Map state functionality with dynamic concurrency adjustment mechanisms that respond to real-time execution metrics through CloudWatch monitoring services. Experimental results demonstrate consistent performance improvements across multiple dimensions, including execution duration, resource utilization efficiency, and operational cost optimization. The adaptive orchestration patterns exhibit enhanced scalability characteristics when processing variable dataset sizes, maintaining stable performance under increasing workload conditions. Implementation considerations encompass architectural patterns for fault tolerance, error handling mechanisms, and comprehensive monitoring frameworks essential for production serverless environments. The findings provide actionable insights for enterprise-scale serverless application development and demonstrate the effectiveness of intelligent orchestration strategies in modern cloud-native architectures.