10 papers
InfraBench: Evaluating Infrastructure Agents Across Layers, Lifecycle, and Risk
Yuan Gao, Zeren Yang, Junnan Li +6
Managing modern computing infrastructure has become a steadily harder problem due to the ever-increasing complexity. Recent advances in AI agents create a timely opportunity to aut…
Don't Let AI Agents YOLO Your Files: Information and Control in Agent-Native Filesystems
Shawn Wanxiang Zhong, Junxuan Liao, Jing Liu +3
AI coding agents operate directly on users' filesystems and regularly corrupt data, delete files, and leak secrets. We conduct the first systematic study of agent filesystem misuse…
TensorHub: Scalable and Elastic Weight Transfer for LLM RL Training
Chenhao Ye, Huaizheng Zhang, Mingcong Han +11
Modern LLM reinforcement learning (RL) workloads require a highly efficient weight transfer system to scale training across heterogeneous computational resources. However, existing…
OBASE: Object-Based Address-Space Engineering to Improve Memory Tiering
Vinay Banakar, Suli Yang, Kan Wu +3
Hardware and OS mechanisms for memory tiering are widely deployed, yet datacenters still overprovision DRAM. The root cause is hotness fragmentation: allocators place objects by si…
Getting the MOST out of your Storage Hierarchy with Mirror-Optimized Storage Tiering
Kaiwei Tu, Kan Wu, Andrea C. Arpaci-Dusseau +1
We present Mirror-Optimized Storage Tiering (MOST), a novel tiering-based approach optimized for modern storage hierarchies. The key idea of MOST is to combine the load balancing a…
Tidying Up the Address Space
Vinay Banakar, Suli Yang, Kan Wu +3
Memory tiering in datacenters does not achieve its full potential due to hotness fragmentation -- the intermingling of hot and cold objects within memory pages. This fragmentation…