Institutional Repository of Vilnius University
Duomenų perbalansavimo našumo palyginimas NoSQL duomenų bazėse /
Abstract
dc:descriptionThis article examines the performance impact of data rebalancing in MongoDB and Apache Cassandra, representing the document and wide-column NoSQL database types, respectively. While there is a fair amount of research analyzing the performance of these two systems under various scenarios, database configurations, and other variables, no academic study was found that treats rebalancing as a separate and important process in a database lifecycle. For this reason, the experiment that was conducted offers a new perspective and adds a lot to the existing literature on NoSQL databases. First, based on previous NoSQL database performance experiments and theoretical material related to rebalancing, a metric set was developed consisting of primary and auxiliary metrics, aimed at evaluating the rebalancing performance of NoSQL systems. The actual experiment was then carried out on the VU MIF high-computing cluster, with the databases and other required tools deployed and run as containers on separate physical cluster nodes. YCSB, a tool that is commonly used for measuring NoSQL database performance, was chosen for workload generation and initial data loading. Both databases, each initially consisting of 6 nodes and loaded with 70 gigabytes of data, were tested under write-heavy and read-heavy workloads during rebalancing. To measure the rebalancing performance of each system, tests were run both in a stable state to measure baseline performance as well as during scale-out by adding 3 new nodes and scale-in by removing previously added nodes. The experiment revealed that rebalancing in MongoDB has less impact on overall system performance, but takes up to 20 times longer during scale-out when compared to Apache Cassandra, where performance drops are more noticeable to the end users. It was also found that both databases perform better during scale-out than scale-in. Finally, based on the obtained results, empirically grounded recommendations were provided for choosing the right database from a rebalancing perspective.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Šaparauskas, Kipras,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD308338819&prefLang=en_US