FastEEP：MoE 推理服务实例内弹性扩缩容系统
FastEEP: Fast Intra-Instance Elastic Scaling for Expert-Parallel MoE Serving
MoE 服务扩容不用再整实例重启了，FastEEP 复用运行时状态，比 vLLM 方案快 4.5 到 11.7 倍，做推理部署的可以看看。
论文提出 FastEEP，一个面向 MoE 模型专家并行（EP）部署的实例内弹性扩缩容系统。现有 vLLM 等推理引擎依赖实例级扩缩容，需要销毁并重建整个服务实例，开销大。FastEEP 基于 Partial Reuse 原则，在 worker 执行、GPU 显存和通信层面复用已有运行时状态，无需完整重建实例。评测显示，FastEEP 的扩缩容速度比基于 vLLM 的方案快 4.5 至 11.7 倍，可实现对 MoE 服务的细粒度弹性伸缩。
FastEEP: Fast Intra-Instance Elastic Scaling for Expert-Parallel MoE Serving
Large-scale expert parallelism (EP) has become the standard deployment configuration for Mixture-of-Experts (MoE) models in production. Meanwhile, the highly dynamic nature of LLM workloads requires providers to continuously adjust serving capacity to match fluctuating demand. Scaling MoE serving cannot be efficiently achieved by simply adding or removing serving instances; instead, it often requires resizing the EP group within an existing instance, which we refer to as intra-instance scaling.Existing inference engines primarily rely on instance-level scaling, which destroys and reconstructs entire serving instances, incurring substantial scaling overhead. We present FastEEP (Fast Elastic Expert Parallel), a fast intra-instance scaling system for MoE serving based on the principle of Partial Reuse. FastEEP reuses existing runtime state across worker execution, GPU memory, and communication by separating reusable execution structures from the resources activated by the current EP configuration. This separation enables coordinated resource reclamation and reconfiguration while preserving costly-to-reconstruct state, allowing fast EP scaling without full instance teardown and reconstruction. Our evaluation shows that FastEEP achieves 4.5x--11.7x faster scaling than the existing vLLM-based approach across different models, enabling efficient, fine-grained scaling of MoE serving.