Design and Implementation of a Strongly Consistent Distributed Key–Value Store (PyazDB)

This paper has employed the Raft consensus algorithm to create and build a strongly consistent distributed key-value store. The primary goal of the work was to prove that consensus-based replication can be applied to correctness and fault tolerance of distributed storage systems. The proposed system provides the ability to ensure that all replicas take note of operations in a well-defined order by having the leader-based architecture and having majority agreement requirement before committing updates.   These findings demonstrate that the system is linearizable in normal operation conditions and it can deal with crash failures without loss to data integrity. The ability to recover automatically whenever there is a failure of any node in the network is made possible by leader election and log replication mechanisms, whereas intentional rejection of operations whenever there is quorum loss maintained safety. Moreover, the separation of consensus logic and application logic followed by its modular design is more compelling to clarity and reliability. All in all, this paper shows that Raft is a useful and comprehensible way of constructing strongly-consistent distributed systems.

  [1]          A. Sharov, A. Shraer, A. Merchant, and M. Stokely, “Take me to your leader! Online Optimization of Distributed Storage Configurations,” Google, Inc. Technical Report, 2015. [2]          R. Batra, “Implementation and Evaluation of Paxos and Raft Distributed Consensus Protocols,” Master’s report, The University of Texas at Austin, 2017. [3]          D. Vishwanath, “Validating Key-value Based Implementations of the Raft Consensus Algorithm for Distributed Systems,” Master’s project, San Jose State University, 2018. [4]          H. Howard and R. Mortier, “Paxos vs Raft: Have we reached consensus on distributed consensus?” in Proceedings of the 7th Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC ’20), 2020. [5]          N. V. Lewchenko and G. Kaki, “Distributed Consensus Algorithms as Replicated State Applications,” in Proceedings of the 10th Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC ’23), University of Colorado Boulder, 2023. [6]          P. S. Almeida, “A Framework for Consistency Models in Distributed Systems: Time, Visibility, Order, Convergence, Basic Axioms, and a New Trilemma,” arXiv preprint arXiv:2411.16355, 2024. [7]          J. Hu, “Distributed Game Server based on RAFT Consensus Core,” Technical Report, University of California, San Diego, 2024. [8]          K. Murugan, “Performance Optimization for etcd using Write-Ahead Logging,” Technical Report, 2024. [9]          N. S. Pasupuleti, “Optimizing Read Performance in Distributed Key-Value Stores Using Serializable Consistency,” International Journal of Innovative Research in Computer and Communication Technology (IJIRCT), vol. 10, no. 3, 2024. [10]        T. Afifi, A. Hegazy, and E. Abousaif, “From Consensus to Chaos: A Vulnerability Assessment of the RAFT Algorithm,” International Journal of Advanced Computer Science and Applications (IJACSA), vol. 16, no. 12, 2025. [11]        C. M. Alves et al., “Composing State Machine Replication,” Technical Paper, 2025. [12]        L. Mastromauro, D. S. Andrade, M. O. Ozmen, and M. A. Kinsy, “Survey of Attacks and Defenses on Consensus Algorithms for Secure Data Replication in Distributed Systems,” IEEE Access, vol. 13, pp. 143644–143667, 2025. [13]        A. Scrivano, “Distributed Database Systems and Consensus Protocols: Understanding the Challenges and Solutions in Achieving Consistency and Reliability,” Review Article, 2025. [14]        M. Shah and A. V. Hazarika, “An In-Depth Analysis of Modern Caching Strategies in Distributed Systems: Implementation Patterns and Performance Implications,” International Journal of Science and Engineering Applications, vol. 14, no. 1, pp. 09–13, 2025. [15]        K. Chen and L. Shi, “Decentralized DHR Architecture for Network Security,” Research Paper, 2026.