MaxFlowSDN: SDN-Based Maximum Flow Routing for High-Throughput Data Center Networks
Large data centers continuously move data from one server to another due to up-or downscaling virtual server specifications to meet user requirements.Most data center topologies allow multipath routing between any pair of nodes within their network to increase throughput, as well as resilience against link failures.Several approaches have been proposed and developed to utilize these routes to improve network performance.Existing methods often face challenges in achieving maximum throughput across diverse topologies without requiring kernel modifications, leaving room for improvement in practicality and scalability.In this paper, we propose a novel system, MaxFlowSDN, which uses the maximum flow algorithm along with traditional SDN and TCP to deliver higher throughput in data centers.MaxFlowSDN yielded 80% higher throughput in the Fat-Tree topology compared to StandardTCP, ParallelTCP, and MPTCP.In DCell and BCube topologies, it achieved approximately 190% higher throughput than StandardTCP and nearly 50% improvement over ParallelTCP and MPTCP.For evaluation, we deployed our system in different data center topologies and compared our results against existing methods.These results demonstrate that MaxFlowSDN provides maximum flow throughput in the data center environment while addressing the limitations of current approaches.
Authors
- Norah S. Bin Saeed
- Maazen Alsabaan (ORCID: https://orcid.org/0000-0001-8601-3184)
- Mohammed J. F. Alenazi (ORCID: https://orcid.org/0000-0001-6593-112X)
- Abdulrahman Almutari
Institutions
- King Saud University (SA)
Publication Details
- Journal
- Ingénierie des systèmes d information
- Published
- 2025-06-30
- DOI
- https://doi.org/10.18280/isi.300611
- Primary Topic
- Software-Defined Networks and 5G
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- King Saud University