Memory-Efficient Key-Value Stores: Evaluating and Extending SILT

Designing storage systems that balance high performance with extremely low memory overhead is a critical challenge in modern systems engineering. Understanding how architectures like SILT (Simple Index Log Structured Table) achieve these goals provides a blueprint for building next-generation key-value stores. This text-only course guides you through the fundamental design principles of memory-efficient key-value stores. You will transition from understanding basic storage concepts to critically evaluating system architectures and designing custom extensions for modern hardware workloads. What you'll learn: Understand the foundational architecture of SILT and its multi-stage index design; Evaluate system performance against memory, write, and read amplification goals; Analyze how log-structured storage optimizes data layout for flash memory; Design architectural extensions to adapt key-value stores for modern workloads; Compare SILT trade-offs with traditional LSM-trees and B-trees. You will begin with core storage terminology and the mechanics of flash-based key-value stores before moving into architectural evaluation frameworks and conceptual exercises on extending the design. This course is designed for software engineers, systems architects, and computer science students new to storage engine internals, requiring no prior database design experience. Start reading today to master the principles of high-efficiency storage system design.