A Comprehensive Analysis of the Murmur3Partitioner in Apache Cassandra: Architecture, Performance, and Implementation Considerations
DOI:
https://doi.org/10.63282/3050-922X.IJERET-V4I4P114Keywords:
Apache Cassandra, Murmur3Partitioner, Consistent Hashing, Data Partitioning, Load Balancing, Distributed Databases, Replication StrategiesAbstract
Partitioning is a foundational mechanism within Apache Cassandra’s distributed database architecture, determining the location of data across the token ring and directly influencing throughput, latency, load balance, and operational resilience. Since Cassandra 1.2, the Murmur3Partitioner has served as the default component for computing partition tokens, replacing earlier strategies that relied either on MD5 hashing or lexicographically ordered keys.[1], [2] Although the Murmur3Partitioner is pervasive in modern Cassandra deployments, the system-level implications of its hashing design, its interactions with consistent hashing, and its long-term effects on cluster behavior remain relatively underexamined in scientific literature[3]. This paper provides a detailed and comprehensive analysis of the Murmur3Partitioner, including its algorithmic foundations in MurmurHash3, its role within Cassandra’s consistent hashing ring[3], its implications for virtual node architecture[3], and its behavior under real-world and adversarial workloads[1]. Extended code samples in Java and Python illustrate practical usage scenarios such as token computation, placement prediction, and debugging data skew[2]. The paper also examines the performance characteristics of Murmur3 under different workload distributions, the operational consequences of token imbalance, and the broader architectural relationships between hashing, replication[2], and compaction. Finally, the paper identifies potential future research directions, including learned partitioners, adaptive hashing systems, and machine-learning-assisted workload prediction, providing a forward-looking perspective on partitioning within large-scale distributed database systems.
References
[1] M. B. Brahim, W. Drira, F. Filali, and N. Hamdi, “Spatial data extension for Cassandra NoSQL database,” Journal Of Big Data , vol. 3, no. 1, Jun. 2016, doi: 10.1186/s40537-016-0045-4.
[2] S. Yamaguchi and Y. MORIMITSU, “Improving Dynamic Scaling Performance of Cassandra,” IEICE Transactions on Information and Systems , no. 4, p. 682, Jan. 2017, doi: 10.1587/transinf.2016dap0009.
[3] H. Chihoub and C. Collet, “A Scalability Comparison Study of Data Management Approaches for Smart Metering Systems,” p. 474, Aug. 2016, doi: 10.1109/icpp.2016.61.
[4] Lewi, K., Kim, W., Maykov, I., & Weis, S. (2019). Securing update propagation with homomorphic hashing. IACR Cryptology ePrint Archive, Report 2019/227. https://eprint.iacr.org/2019/227
[5] S. P. Kumar, “Adaptive Consistency Protocols for Replicated Data in Modern Storage Systems with a High Degree of Elasticity,” HAL (Le Centre pour la Communication Scientifique Directe) , Mar. 2016, Accessed: Sep. 2025. [Online]. Available: https://tel.archives-ouvertes.fr/tel-01359621
[6] T. Rabl, S. Gómez-Villamor, M. Sadoghi, V. Muntés-Mulero, H. Jacobsen, and S. Mankovskii, “Solving big data challenges for enterprise application performance management,” Proceedings of the VLDB Endowment , vol. 5, no. 12, p. 1724, Aug. 2012, doi: 10.14778/2367502.2367512.
[7] K. Shankar, A. Mahgoub, Z. Zhou, U. Priyam, and S. Chaterji, “Asgard: Are NoSQL databases suitable for ephemeral data in serverless workloads?,” Frontiers in High Performance Computing , vol. 1, Sep. 2023, doi: 10.3389/fhpcp.2023.1127883.
[8] В. Нікітін and Є. Крилов, “Алгоритм хешування з підвищеною колізійною стійкістю для підтримки консистентності в розподілених базах даних,” Адаптивні системи автоматичного управління , vol. 2, no. 41, p. 45, Dec. 2022, doi: 10.20535/1560-8956.41.2022.271338.
[9] Dabbagh, M., Hamdaoui, B., Guizani, M., & Rayes, A. (2015). Energy-Efficient Resource Allocation and Provisioning Framework for Cloud Data Centers. IEEE Transactions on Network and Service Management, 12(3), 377–391. https://doi.org/10.1109/TNSM.2015.2436408
[10] Cohen, E., Delling, D., Pajor, T., & Werneck, R. F. (2014). Sketch-based influence maximization and computation: Scaling up with guarantees. Proceedings of the 23rd ACM International Conference on Conference on Information and Knowledge Management (CIKM ’14), 629-638. https://doi.org/10.1145/2661829.2662015
[11] C. Fu, O. Bian, H. Jiang, L. Ge, and H. Ma, “A New Chaos-based Image Cipher Using a Hash Function,” The International journal of networked and distributed computing , vol. 5, no. 1, p. 37, Dec. 2016, doi: 10.2991/ijndc.2017.5.1.4.
[12] G. DeCandia et al. , “Dynamo,” p. 205, Oct. 2007, doi: 10.1145/1294261.1294281.
[13] J. S. Filho, D. M. Cavalcante, L. O. Moreira, and J. C. Machado, “An adaptive replica placement approach for distributed key‐value stores,”Concurrency and Computation Practice and Experience, vol. 32, no. 11, Feb. 2020, doi: 10.1002/cpe.5675.
[14] E. A. Khashan, A. I. El-Desouky, and S. M. Elghamrawy, “An adaptive spark-based framework for querying large-scale NoSQL and relational databases,” PLoS ONE , vol. 16, no. 8, Aug. 2021, doi: 10.1371/journal.pone.0255562.
[15] Chain, P., & Myers, E. W. (2007). Comparative genomics: Genome structure, function, and evolution. BMC Genomics, 8(Suppl 2), S3. https://doi.org/10.1186/1471-2164-8-S2-S3
[16] Z. Peng and B. Plale, “Reliable access to massive restricted texts: Experience‐based evaluation,” Concurrency and Computation Practice and Experience , vol. 32, no. 16, Apr. 2019, doi: 10.1002/cpe.5255.
[17] S. Ghule and R. Vadali, “A review of NoSQL Databases and Performance Testing of Cassandra over single and multiple nodes,”Annals of Computer Science and Information Systems , vol. 10. Polskie Towarzystwo Informatyczne, p. 33, Jun. 09, 2017. doi: 10.15439/2017r65.
[18] A. Nanjappan, “R*-Tree index in Cassandra for Geospatial Processing,” 2019. doi: 10.31979/etd.55t5-e77a.
[19] Y. Liu, D. Gureya, A. Al-Shishtawy, and V. Vlassov, “OnlineElastMan: self-trained proactive elasticity manager for cloud-based storage services,” Cluster Computing , vol. 20, no. 3, p. 1977, May 2017, doi: 10.1007/s10586-017-0899-z.
[20] K. Bohora, A. Bothe, D. Sheth, and R. Chopade, “Backup and Recovery Mechanisms of Cassandra Database: A Review,” The journal of digital forensics, security and law . Association of Digital Forensics, Security and Law, Jan. 01, 2021. doi: 10.15394/jdfsl.2021.1613.
[21] Rana, I. A., Ali, R., & Khan, M. M. (2018). Analysis of query optimization components in distributed database. Indian Journal of Science and Technology, 11(18), 1–10.
[22] M. Diogo, B. Cabral, and J. Bernardino, “CBench-Dynamo: A Consistency Benchmark for NoSQL Database Systems,” in Lecture notes in computer science , Springer Science+Business Media, 2020, p. 84. doi: 10.1007/978-3-030-55024-0_6.
[23] K. Grolinger, W. A. Higashino, A. Tiwari, and M. A. M. Capretz, “Data management in cloud environments: NoSQL and NewSQL data stores,” Journal of Cloud Computing Advances Systems and Applications , vol. 2, no. 1, Dec. 2013, doi: 10.1186/2192-113x-2-22.
[24] R. T. Venkatesh, D. K. Chandrashekar, P. B. S. Rao, R. Sridhar, and R. Sunitha, “Systematic Approaches to Data Placement, Replication and Migration in Heterogeneous Edge-Cloud Computing Systems: A Comprehensive Literature Review,” Ingénierie des systèmes d information , vol. 28, no. 3, p. 751, Jun. 2023, doi: 10.18280/isi.280326.
[25] R. Vilaça, R. Oliveira, and J. Pereira, “A Correlation-Aware Data Placement Strategy for Key-Value Stores,” in Lecture notes in computer science , Springer Science+Business Media, 2011, p. 214. doi: 10.1007/978-3-642-21387-8_17.
[26] Vaishya, A., Chandramouli, A., Kale, S., & Krishnan, P. (2022). Coded data rebalancing for distributed data storage systems with cyclic storage. arXiv. https://doi.org/10.48550/arXiv.2205.06257
[27] D. Vasilas, “A flexible and decentralised approach to query processing for geo-distributed data systems,” HAL (Le Centre pour la Communication Scientifique Directe) , Feb. 2021, Accessed: Apr. 2025. [Online]. Available: https://hal.inria.fr/tel-03272208
[28] Ashkiani, S., Farach-Colton, M., & Owens, J. D. (2018). A dynamic hash table for the GPU. In Proceedings of the 2018 IEEE International Parallel and Distributed Processing Symposium (IPDPS) (pp. 419–429). IEEE. https://doi.org/10.1109/IPDPS.2018.00052
[29] Roy, P., Seshadri, S., Sudarshan, S., & Bhobe, S. (1999). Efficient and extensible algorithms for multi-query optimization. Proceedings of the 1999 ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery. https://arxiv.org/abs/cs/9910021
[30] A. A. H. Al-Fatlawi, G. N. Mohammed, and I. A. Barazanchi, “Optimizing the Performance of Clouds Using Hash Codes in Apache Hadoop and Spark,”Journal of Southwest Jiaotong University , vol. 54, no. 6, Jan. 2019, doi: 10.35741/issn.0258-2724.54.6.3.
[31] Abdelhafiz, B. M. (2020). Distributed database using sharding database architecture. In 2020 IEEE Asia-Pacific Conference on Computer Science and Data Engineering (CSDE) (pp. 1–17). IEEE. https://doi.org/10.1109/CSDE50874.2020.9411547
[32] M. Coluzzi, A. Brocco, and A. Antonucci, “BinomialHash: A Constant Time, Minimal Memory Consistent Hash Algorithm,”arXiv (Cornell University) , Jun. 2024, doi: 10.48550/arxiv.2406.19836.
[33] O. Stetsyk and S. Terenchuk, “COMPARATIVE ANALYSIS OF NOSQL DATABASES ARCHITECTURE,” Management of Development of Complex Systems , no. 47, p. 78, Sep. 2021, doi: 10.32347/2412-9933.2021.47.78-82.
[34] Malkowski, S., Hedwig, M., Jayasinghe, D., Park, J., Kanemasa, Y., & Pu, C. (2009). A new perspective on experimental analysis of N-tier systems: Evaluating database scalability, multi-bottlenecks, and economical operation. In Proceedings of the 5th International ICST Conference on Collaborative Computing: Networking, Applications, Worksharing (pp. 1–10). IEEE/ICST. https://doi.org/10.4108/ICST.COLLABORATECOM2009.8311
[35] Pietzuch, P. R., & Bacon, J. M. (2002). Hermes: A distributed event-based middleware architecture. In Proceedings of the Workshop on Distributed Event-Based Systems (DEBS 2002). ACM.
[36] S. A. M. Ariff, S. Azri, U. Ujang, and T. L. Choon, “ORGANIZING SMART CITY DATA BASED ON 3D POINT CLOUD IN UNSTRUCTURED DATABASE – AN OVERVIEW,” The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences , p. 87, Dec. 2022, doi: 10.5194/isprs-archives-xlviii-4-w3-2022-87-2022.
[37] Monnerat, L. R., & Amorim, C. L. (2014). An effective single-hop distributed hash table with high lookup performance and low traffic overhead. Concurrency and Computation: Practice and Experience, 27(15), 3880–3901. https://doi.org/10.1002/cpe.3342
[38] S. M. Elghamrawy, “An Adaptive Load-Balanced Partitioning Module in Cassandra Using Rendezvous Hashing,” in Advances in intelligent systems and computing , Springer Nature, 2016, p. 587. doi: 10.1007/978-3-319-48308-5_56.
[39] A. Davoudian, L. Chen, H. Tu, and M. Liu, “A Workload-Adaptive Streaming Partitioner for Distributed Graph Stores,” Data Science and Engineering , vol. 6, no. 2, p. 163, Apr. 2021, doi: 10.1007/s41019-021-00156-2.
[40] M. A. U. Nasir, G. D. F. Morales, D. García-Soriano, N. Kourtellis, and M. Serafini, “The power of both choices: Practical load balancing for distributed stream processing engines,” p. 137, Apr. 2015, doi: 10.1109/icde.2015.7113279.
