Authors: Veronika Abramova; Jorge Bernardino; Pedro Furtado
Addresses: FCTUC – University of Coimbra, 3030-290 Coimbra, Portugal ' Polytechnic Institute of Coimbra, 3030-190 Coimbra, Portugal ' FCTUC – University of Coimbra, 3030-290 Coimbra, Portugal
Abstract: The big data era brought a set of new database features, such as parallel execution of requests and data distribution among different nodes as well as new types of databases. NoSQL technology emerged to aid people and companies to easily scale systems with simple and transparent data distribution. It became possible to cope with higher demand in less time while performing different types of operations and storing large amounts of data. In this paper, we evaluate Cassandra's scalability and execution time of CRUD operations and, posteriorly, compare one relational and one non-relational system by evaluating their performance during execution of decision support queries. For that purpose, we used two standard benchmarks, Yahoo! Cloud Serving Benchmark, to evaluate execution time of requests and speedup of Cassandra, and Star-Schema Benchmark, to run queries over MySQL cluster, as relational database, and Hadoop with Hive as non-SQL counterpart. We conclude about the capabilities and limitations of those systems.
Keywords: big data; benchmarking; structured query language; SQL; NoSQL; Hadoop; Hive; decision support queries; performance evaluation; scalability evaluation; Yahoo! Cloud Serving Benchmark; YCSB; Cassandra; MySQL cluster; execution time; CRUD operations; relational databases.
International Journal of Business Process Integration and Management, 2015 Vol.7 No.4, pp.314 - 321
Published online: 15 Dec 2015 *Full-text access for editors Access for subscribers Purchase this article Comment on this article