Abstract:Software Defined Networks (SDNs) offer a comprehensive network view by separating the control plane from the data plane. However, SDNs are vulnerable to Distributed Denial of Service (DDoS), a dangerous attack that depletes resources, preventing service delivery. Among the DDoS attacks, the HTTP slow-rate DDoS attack is particularly critical, targeting web servers with slow or incomplete requests. Significant efforts have been made in the last few years to improve DDoS attack detection in SDNs, leading to the proposal of several detection techniques. In an effort to address these current constraints, scientists have concentrated on leveraging the computational capabilities of data plane devices. Notably, in this context, Programming Protocol-independent Packet Processors (P4) have become an important technology closely linked to the data plane components of SDN. The use of new detection techniques through the use of P4-equipped data planes for DDoS detection methods has the potential to reduce the computational load on the controller. This research paper analyzes detection system components and introduces P4httpGuard,a detection mechanism that employs machine learning (ML) techniques in conjunction with P4 switches to identify slow-rate DDoS attacks within SDNs. The model uses P4 switches programmable capabilities to enhance detection while reducing controller computational overhead. The model has been evaluated for performance metrics like detection time, bandwidth consumption, and CPU usage. The results from the implementation of our mechanism demonstrate a notable 60-second improvement in detection time, an 81.89% reduction in bandwidth consumption, and a 25.96% decrease in controller CPU overhead, in compare to the Openflow method. These findings underscore the significant impact of integrating the P4 data plane and programmable targets in substantially enhancing the efficiency of slow-rate DDoS attack detection within SDN.

Load Profiling via In-Band Flow Classification and P4 With Howdah

An Efficient Sampling And Classification Approach For Flow Detection In Sdn-Based Big Data Centers

Flow Anomaly Telemetry Driven by Programmable Data Plane

Unleashing Dynamic Pipeline Reconfiguration of P4 Switches for Efficient Network Monitoring

Performance Modeling and Analysis of P4 Programmable Devices With General Service Times

Advancements in Traffic Processing Using Programmable Hardware Flow Offload

Accelerating End-host Congestion Response using P4 Programmable Switches

P4httpGuard: detection and prevention of slow-rate DDoS attacks using machine learning techniques in P4 switch

P8: P4 With Predictable Packet Processing Performance

Hedera: Dynamic Flow Scheduling for Data Center Networks

An Efficient Flow Detection and Scheduling Method in Data Center Networks

Toward Low-Overhead Inter-Switch Coordination in Network-Wide Data Plane Program Deployment

Introducing Packet-Level Analysis in Programmable Data Planes to Advance Network Intrusion Detection

Research on Traffic Identification and Scheduling Based on Optical Interconnection Architecture in Data Center

An Efficient Design of Intelligent Network Data Plane

Freeway: Adaptively Isolating the Elephant and Mice Flows on Different Transmission Paths

Rethinking Transfer Optimization in a Datacenter: Integrating Load Balancing with Multipath Flow Control.

FlowDyn: Towards a Dynamic Flowlet Gap Detection using Programmable Data Planes

P4DM: Measure the Link Delay with P4