COMPARATIVE ANALYSIS OF VERTICAL AND HORIZONTAL SCALING METHODS IN COMPUTING SYSTEMS

DOI: 10.31673/2412-4338.2024.036268

Abstract

The article addresses the pressing issue of scaling computing systems, which is a key factor in ensuring the efficient operation of modern information systems. The research focuses on a comparative analysis of two main approaches to scaling: vertical and horizontal. The aim of the study is to determine the optimal scaling method for different types of workloads and computing resources. To achieve this, a detailed analysis of existing research was conducted, along with experimental studies based on models of computing systems. The study found that the choice of scaling method depends on specific system requirements such as cost, performance, reliability, and flexibility. Another important criterion is the availability of specialists and infrastructure for implementing a particular solution, which affects the overall choice. It was found that horizontal scaling is more effective for systems with highly dynamic workloads, while vertical scaling is better suited for systems with high computational power requirements for individual nodes. The results obtained can be used to develop recommendations for choosing the optimal scaling method for specific applications. The findings were derived using existing IaaS (Infrastructure as a Service) platforms like AWS and Azure, as well as deploying a test server in local conditions. Future research prospects include the development of hybrid scaling methods that combine the advantages of both vertical and horizontal scaling, as well as studying the impact of new technologies such as containerization and serverless architectures on scaling efficiency.

Keywords: scaling, vertical scaling, horizontal scaling, computing systems, reliability.

References
1. A. Kwan, J. Wong, H.A. Jacobsen and V. Muthusamy, "HyScale: Hybrid and Network Scaling of Dockerized Microservices in Cloud Data Centres," 2019 IEEE 39th International Conference on Distributed Computing Systems (ICDCS), Dallas, TX, USA, 2019, pp. 80-90
2. Afzal, S., Kavitha, G. Load balancing in cloud computing – A hierarchical taxonomical classification. J Cloud Comp 8, 22 (2019).
3. S. Spinner et al., "Runtime Vertical Scaling of Virtualized Applications via Online Model Estimation," 2014 IEEE Eighth International Conference on Self-Adaptive and Self-Organizing Systems, London, UK, 2014, pp. 157-166
4. What are race conditions?: Some issues and formalizations 1992. URL: https://dl.acm.org/doi/abs/10.1145/130616.130623
5. Bhagwan, R., Savage, S., Voelker, G.M. (2003). Understanding Availability. In: Kaashoek, M.F., Stoica, I. (eds) Peer-to-Peer Systems II. IPTPS 2003. Lecture Notes in Computer Science, vol 2735. Springer, Berlin, Heidelberg.
6. Sivasubramanian S. Szymaniak M. Pierre G. van Steen M. (2004). Replication for web hosting systems.ACM Computing Surveys, 36(3), 291–334.
7. Sakurai, H., Phung-Duc, T. Scaling limits for single server retrial queues with two-way communication. Ann Oper Res 247, 229–256 (2016).