DeltaBox: Millisecond-Scale Sandbox Checkpoint and Rollback for Stateful AI Agents
cs.OS, cs.AI
Submitted: 2026-05-21
Updated: 2026-11-13
Comments: 16 pages, 9 figures, 3 tables. Accepted to SIGOPS ATC 2026
Code: https://github.com/openclaw/openclaw
Project page: https://openzfs.github.io/openzfs-docs
License: http://creativecommons.org/licenses/by/4.0/
The gist: Emerging LLM-powered AI agent workloads, such as test-time tree search and reinforcement learning (RL), require high-frequency state exploration…
Terminology
Abstract
Emerging LLM-powered AI agent workloads, such as test-time tree search and reinforcement learning (RL), require high-frequency state exploration, relying on rapid checkpoint/restore (C/R) of the complete sandbox state, including files and process state. Existing mechanisms duplicate the entire state, causing hundreds of milliseconds to seconds of latency per C/R, which severely bottlenecks deep search and large-scale fan-outs. We observe that consecutive sandbox states between checkpoints differ only marginally. Therefore, instead of full duplication, a sandbox should only duplicate the changes between consecutive checkpoints (Key Insight), yet the lack of OS support makes this hard to realize. We propose a new OS-level abstraction, DeltaState, that delivers change-based C/R for agents through co-designed OS mechanisms. First, DeltaFS checkpoints file state by freezing the writable layer and inserting a new one, turning updates into copy-on-write and rollback into a layer switch. Second, DeltaCR quickly restores process state by forking from a frozen template process. A coordinator pairs the two so each checkpoint is one consistent (filesystem, memory) state. Building on these mechanisms, we present DeltaBox, an agent sandbox with millisecond-level C/R. On SWE-bench, DeltaBox hides its checkpoint work (10.83 ms) under inference and rolls back in a mean 1.86 ms, with the same warm-template fork driving low-cost RL rollout fan-out. DeltaBox is open-sourced at https://github.com/delta-box.
Sources
- MobiAgent: A Systematic Framework for Customizable Mobile Agents
- Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Training Long-Context, Multi-Turn Software Engineering Agents with Reinforcement Learning
- OpenRLHF: An Easy-to-use, Scalable and High-performance RLHF Framework
- Fault-Tolerant Sandboxing for AI Coding Agents: A Transactional Approach to Safe Autonomous Execution
- Toward Systems Foundations for Agentic Exploration
- Crab: A Semantics-Aware Checkpoint/Restore Runtime for Agent Sandboxes
- Taming Serverless Cold Starts Through OS Co-Design