DeepSeek DSec: Running 380K Sandboxes with 50x Overcommit for Agent Training

DeepSeek's DSec infrastructure supports millions of agent sandboxes through layered environments, on-demand image loading via 3FS, memory sharing with virtio-pmem/DAX, CPU scheduling, and trajectory forking, achieving 50x resource overcommit while addressing security challenges like agent-discovered vulnerabilities.

3FSAgent TrainingAppArmor

0 likes · 12 min read

DeepSeek DSec: Running 380K Sandboxes with 50x Overcommit for Agent Training

Machine Heart

Sep 23, 2026 · Backend Development

DeepSeek DSec: Scaling Sandbox Infrastructure to 380K Concurrent Agents

DeepSeek's new paper introduces DSec, a production sandbox infrastructure that manages 300 million daily sandbox instances with 380,000 concurrent at peak, using on-demand image loading, composable layers, memory sharing, and CPU QoS to support large-scale agentic reinforcement learning training.

CPU schedulingContainer OrchestrationDSec

0 likes · 19 min read

DeepSeek DSec: Scaling Sandbox Infrastructure to 380K Concurrent Agents