Sarcouncil Journal of Engineering and Computer Sciences

Abstract: Enterprise data architectures have evolved significantly from traditional data warehouses to more flexible data lakes and the emerging lakehouse paradigm. Data warehouses provide structured repositories optimized for business intelligence through dimensional modeling, schema enforcement, and query optimization, but face challenges with cost scalability and unstructured data processing. Data lakes offer cost-efficient storage for diverse data types without predefined schemas, enabling AI/ML workloads while presenting governance and performance challenges. The lakehouse paradigm bridges these approaches by implementing warehouse-like capabilities directly on low-cost cloud storage through technologies such as Delta Lake, Apache Iceberg, and Apache Hudi. This architectural convergence eliminates data duplication while supporting both structured analytics and machine learning workloads against a unified repository. Organizations must evaluate performance characteristics, total cost of ownership, engineering complexity, and scalability requirements when selecting appropriate architectures, with implementation strategies varying based on existing investments and analytical priorities.