Comprehensive Study of Persistence Techniques in In-memory Databases
Aryan Chaudhari, Harsh Bhat, Abhishek Belgaonkar, Aditya Supare, Seema Patil · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2025
In recent years, NoSQL databases have become essential for delivering Big data web services.As memory capacities increase, there is a heightened focus on In-memory NoSQL (IM-NoSQL) systems, which use dynamic random-access memory (DRAM) to enable minimal latency.However, because DRAM is volatile, IM-NoSQL systems need effective persistence and recovery methods to prevent data loss during server failures.This paper presents a detailed study of the performance of persistence and recovery techniques in IM-NoSQL databases.The evaluation examines the performance of Snapshotting and logging techniques, focusing on their effectiveness in failure recovery.Our research aims to answer critical questions: (i) This study investigates whether IM-NoSQL systems maintain efficiency under memory constraints.(ii)What are the performance trade-offs between using snapshots and logging?(iii) How quickly can an IM-NoSQL system recover after a failure?(iv)How does the choice of persistence method affect the system's performance?This study utilizes Redis as a representative IM-NoSQL system to evaluate persistence strategies, recovery durations, and system performance metrics.