一句话：将 DPU SoC 与 Host 上拆分的 hypervisor 视为冗余对等体，让虚拟机从故障节点逃逸并在新机器上恢复。
现代云服务器利用 DPU 卸载虚拟化开销，代价是 hypervisor 功能被拆分到 DPU 与 Host 两侧，机器从单一系统变为多个自治处理单元的编排：任一侧出错都会导致整机崩溃，故障率大幅上升，现有虚拟机容错与隔离机制亟需重构。本文提出故障逃逸（fault escaping）方案，把 DPU SoC 与 Host 上拆分的 hypervisor 视为互为冗余的对等体，借助混合同步以 KB 级开销使完整 VM 状态在两侧均可访问；故障发生时不再修复失效的 hypervisor，而是迁移经过校验的 VM 状态，让虚拟机逃离故障并在新机器上恢复。评估显示该机制不引入运行时性能开销，性能与虚拟机实时迁移相当，并在大型云生产环境中将 VM 恢复时间目标（RTO）缩短 4～14 倍。
Modern cloud servers achieve significant performance improvements by exploiting data processing units (DPUs) to offload virtualization overhead. Unlike traditional monolithic hypervisors, this offloading approach splits VM state and distributes hypervisor functions across the DPU and the Host, transforming the cloud server from a single system into a sophisticated orchestration of multiple self-managed processing units. However, the failure rate of these systems has significantly increased, as errors from either the DPU or the Host can crash the entire machine. All these changes necessitate a comprehensive revamp of current VM fault tolerance and isolation mechanisms. This paper explores a novel approach to VM fault tolerance by treating the split hypervisors on the DPU SoC and the Host as redundant peers. We propose a fault-escaping scheme that enables VMs to escape from a failing SoC or Host. With hybrid synchronization, the entire VM state becomes fully accessible on either the SoC or the Host with minimum synchronization overhead (in kilobytes). Instead of recovering the failed hypervisor, this scheme migrates the verified VM state, allowing the VM to escape from the failure and resume on a new machine. The evaluation demonstrates that the escape-based mechanism introduces no runtime performance overhead, achieves performance levels comparable to VM live migration, and reduces the VM recovery time objective (RTO) by 4x to 14x in one of the largest cloud production environments.
@inproceedings{Zhang0LLXYSLLLS25,
title = {Fault Escaping: Improving Robustness of DPU Enhanced Platform with Mutual Assisted VM Recovery},
author = {Chao Zhang and Tao Xu and Junming Liu and Pai Liu and Zhilang Xu and Kang Yan and Shuo Shi and Jintao Li and Jinhu Li and Chen Liang and Wenhui Shu and FeiFei Fan and Yibin Shen and Hang Yang and Jianming Song and Xudong Zheng and Jiesheng Wu and Jian Li},
booktitle = {ASPLOS},
year = {2025},
doi = {10.1145/3676642.3736124}
}