Year: 2024
Type: article
Abstract: Java-based big data applications are widely used to process and analyze massive datasets in distributed environments, powering critical systems in finance, healthcare, e-commerce, and cloud computing. However, achieving optimal performance in these applications remains a persistent challenge due to JVM overhead, memory constraints, data skew, network latency, and inefficient resource utilization. This study investigates performance optimization techniques specifically tailored for Java-driven big data platforms such as Apache Hadoop, Spark, and Flink. The research aims to identify architectural, algorithmic, and runtime-level strategies that improve execution efficiency and scalability. The methodology combines a systematic review of existing literature with experimental evaluation of optimization techniques, including memory tuning, garbage collection optimization, parallel processing strategies, data partitioning, caching mechanisms, and JVM configuration enhancements. Additionally, workload-aware scheduling and adaptive resource management approaches are analyzed to understand their impact on distributed performance. The findings demonstrate that performance gains are best achieved through a multi-layer optimization strategy that integrates code-level improvements with system-level tuning and intelligent workload management. The study concludes that effective performance optimization in Java-based big data applications requires a holistic approach that balances computation, memory, and I/O efficiency while maintaining system reliability. These insights provide practical guidance for developers and system architects seeking to design high-performance, scalable big data solutions.
Author: John Owen
Institution: Ladoke Akintola University of Technology
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7128468604