KERNEL-ACCELERATED STATELESS LOAD DISTRIBUTION ENGINE WITH EBPF
Dr.K.E.Kannammal Department of Computer Science
and Engineering
Sri Shakthi Institute of Engineering and Technology
Coimbatore, India hodcse@siet.ac.in
Sharveshwaran S S Department of Computer Science
and Engineering
Sri Shakthi Institute of Engineering and Technology
Coimbatore, India sharveshwaranss22cse@srishakthi.ac.in
Soorya Akilesh C Department of Computer Science
and Engineering
Sri Shakthi Institute of Engineering and Technology
Coimbatore, India sooryaakileshc22cse@srishakthi.ac.in
Abstract — The Kernel-Accelerated Stateless Load Distribution Engine with eBPF is a high-performance networking solution designed to meet the demands of modern cloud-native environments requiring ultra-low latency and high throughput. This system implements a Layer-3 load balancer using advanced technologies such as eBPF (extended Berkeley Packet Filter) and XDP (eXpress Data Path), enabling packet processing directly within the Linux kernel at the earliest stage of the network stack, specifically at the network interface driver level. By leveraging kernel-space execution, the system achieves near line-rate performance while minimizing overhead associated with traditional user-space load balancers.
The engine efficiently intercepts incoming IPv4 traffic, applies a hash-based distribution algorithm to ensure stateless and balanced request routing, and rewrites packet headers for seamless forwarding to backend servers. Additionally, the system integrates a real-time web-based dashboard that provides visibility into kernel-level operations, including active eBPF programs, map states, and packet flow metrics. This bridges the gap between low-level kernel networking and high-level system administration, offering both performance and observability. The solution is highly scalable, efficient, and suitable for next-generation distributed systems and cloud infrastructures.
Keywords — eBPF, XDP, Load Balancer, Kernel Networking, High Throughput, Low Latency, Stateless Architecture, Packet Processing, Cloud-Native Systems, Real-Time Monitoring.
