Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper investigates the security vulnerabilities introduced by Remote Direct Memory Access (RDMA) in untrusted environments due to its CPU-bypass design. To address this, the authors implement AES-128 encryption directly within the data plane of a programmable Tofino switch using P4, offloading encryption from the CPU. Experimental results demonstrate the feasibility of this approach, achieving up to 1.9 Gbps throughput for 128-byte packets, thereby preserving RDMA's performance benefits while enhancing security.
Securing RDMA transfers doesn't have to kill performance: offloading AES-128 encryption to a programmable switch achieves up to 1.9 Gbps throughput.
Remote Direct Memory Access (RDMA) is a key enabler of high-performance systems, offering low latency, high throughput, and reduced CPU overhead by allowing direct memory-to-memory transfers between machines. However, its design bypasses traditional CPU-mediated security mechanisms, introducing critical vulnerabilities in untrusted environments. This work explores the integration of RDMA and AES-128 encryption to secure data transfers without compromising performance. We implement encryption directly within the data plane of a programmable Tofino switch using the P4 programming language. By offloading encryption from the CPU to the switch, our design preserves RDMA's performance benefits while addressing its security shortcomings. Experimental results show that the system achieves throughput of 0.37 Gbps for 16-byte packets, 0.76 Gbps for 32-byte packets, 1.83 Gbps for 64-byte packets, and 1.9 Gbps for 128-byte packets. These findings demonstrate the feasibility of secure, high-throughput RDMA communication using programmable network hardware.