Sponsored By



In Cooperation With

Supporters

Gold Sponsors

Futurewei

purestorage

samsung

Bronze Sponsors

DELL Technologies

google

Follow us



Logo design by Michal O. Zadok.



Full Program

Day 1, Thursday, July 10, 2025 (Full Day)

08:00 – 09:00 EDTBreakfast
08:30 – 09:00 EDTRegistration
09:00 – 09:15 EDTOpening Remarks
09:15 – 10:15 EDTKeynote Address by Xiaosong Ma
10:15 – 10:30 EDTCoffee Break
10:30 – 11:45 EDTSession 1: Cloud and Disaggregated Storage I
11:45 – 13:20 EDTLunch
13:20 – 15:00 EDTSession 2: Memory Centric Storage
15:00 – 15:15 EDTCoffee Break
15:15 – 16:05 EDTSession 3: High-Performance Storage I/O
16:05 – 16:20 EDTCoffee Break
16:20 – 17:35 EDTSession 4: Caching and Deduplication
17:35 --20:30 EDTDinner

Day 2, Friday, July 11, 2025 (Half Day)

08:00 – 09:00 EDTBreakfast
09:00 – 09:50 EDTSession 5: Cloud and Disaggregated Storage II
09:50 – 10:00 EDTCoffee Break
10:00 – 11:15 EDTSession 6: Storage and Machine Learning
11:15 – 11:30 EDTCoffee Break
11:30 – 12:20 EDTSession 7: System Overheads and Security
12:20 – 12:30 EDTConcluding Remarks
12:30 EDTLunch (box lunch)

Keynote Talks

I/O Coordination for Better Resource Sharing -- From HPC to AI Storage
Xiaosong Ma (Mohamed bin Zayed University of Artificial Intelligence)
[Abstract], [Speaker Bio], [Slides]
Abstract:

Despite tremendous improvement in absolute capacity and speed, the storage subsystem remains one of the slower, less predictable, and less scalable components of large-scale applications. In this talk, through a personal journey of parallel and distributed storage systems, I hope to share observations and lessons from these past projects. In particular, across the many layers of the storage hierarchy studied, from the CPU caches to supercomputer and cloud storage clusters, we repeatedly encounter the same underlying challenge of efficient sharing of I/O resources. Without effective regulation of I/O flows, the storage system can easily be a performance bottleneck while remaining largely underutilized. The discussion of solutions continues into related new (and old) storage problems with today's parallel AI training and inference workloads.

Bio:

Professor Ma’s main research focus has been on efficient resource utilization across different system layers, from CPU/GPU caches to datacenter-scale distributed storage. Her research projects share the common theme of reducing or eliminating waste in systems software, to better support applications including graph, database, and LLM processing.



Sessions

Session 1: Cloud and Disaggregated Storage I

SnapBPF: Exploiting eBPF for Serverless Snapshot Prefetching
[Paper] [Slides]
Stratos Psomadakis, Dimitrios Siakavaras (National Technical University of Athens); Chloe Alverti (University of Illinois Urbana-Champaign); Symeon Porgiotis, Orestis Lagkas Nikolos, Christos Katsakioris, Konstantinos Nikas, Georgios Goumas, Nectarios Koziris (National Technical University of Athens)

Can a Client–Server Cache Tango Accelerate Disaggregated Storage?
[Paper] [Slides]
Linjie Ma (Rutgers university); Jian Zhang (Rutgers University); Marie Nguyen (Samsung); Sudarsun Kannan (Rutgers University)

Why Paying for Storage Beats Free Networking in Cloud Bursting
[Paper] [Slides]
Itamar Gefen, Aviad Zuck, Daniel Bransky (Technion – Israel Institute of Technology); Moshik Hershcovitch (Tel Aviv University & IBM Research); Danny Harnik (IBM Research); Dan Tsafrir (Technion – Israel Institute of Technology)


Session 2: Memory Centric Storage

Efficient Vector Search on Disaggregated Memory with d-HNSW
[Paper] [Slides]
Yi Liu, Fei Fang, Chen Qian (University of California Santa Cruz)

Revisiting Trim for CXL Memory
[Paper] [Slides]
Hayan Lee (Soongsil University); Jungwoo Kim (DGIST); Wookyung Lee (Soongsil University); Juhyung Park (DGIST); Sanghyuk Jung, Jinki Han (Eeum); Bryan S. Kim (Syracuse University); Sungjin Lee (POSTECH); Eunji Lee (Soongsil University)

Revisiting Memory Hierarchies with CMM-H: Using Device-side Caching to Integrate DRAM and SSD for a Hybrid CXL Memory
[Paper] [Slides]
Mohammadreza Soltaniyeh; Gongjin Sun; Xuebin Yao; Amir Beygi; Ramdas Kachare; Dongwan Zhao; Hingkwan Huen; Andrew Chang; Senthil Murugesapandian; Caroline Kahn (Samsung Semiconductor)

Towards Rack-as-a-Computer in Memory Interconnect Era with Coordinated Operating System Sharing
[Paper] [Slides]
Yuxin Ren, Mingrui Liu, Hongbo Li, Chang Liao, Xiaojia Huang, Jianhua Zhang, Hanjun Guo, Yubo Liu, Ning Jia (Huawei Technologies)


Session 3: High-Performance Storage I/O

SPDK+: Low Latency or High Power Efficiency? We Take Both
[Paper] [Slides]
Endian Li, Shushu Yi, Li Peng (Peking University); Qiao Li (Xiamen University); Diyu Zhou (Peking University); Zhenlin Wang (Michigan Tech); Xiaolin Wang (Peking University); Bo Mao (Xiamen University); Yingwei Luo (Peking University); Ke Zhou (Huazhong University of Science and Technology (HUST)); Jie Zhang (Peking University)

ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe
[Paper] [Slides]
Junhyeok Park (Sogang University); Junghee Lee (Korea University); Youngjae Kim (Sogang University)


Session 4: Caching and Deduplication

CableCache: In-Network Request Deduplication for Key-Value Stores
[Paper] [Slides]
Jiawei Huang, Junru Li, Qing Wang, Lijie Wen, Youyou Lu (Tsinghua University); Erci Xu (SJTU)

PaperCache: In-Memory Caching with Dynamic Eviction Policies
[Paper] [Slides]
Kia Shakiba, Michael Stumm (University of Toronto)

Unlocking the Unusable: A Proactive Caching Framework for Reusing Partial Overlapped Data
[Paper] [Slides]
Chang Guo (Arizona State University); Norbert Podhorszki (Oak Ridge National Laboratory); Greg Eisenhauer (Georgia Institute of Technology); Zhiwen Xie (Arizona State University); Scott Klasky (Oak Ridge National Laboratory); Zhichao Cao (Arizona State University)


Session 5: Cloud and Disaggregated Storage II

Are Edge MicroDCs Equipped to Tackle Memory Contention?
[Paper] [Slides]
Long Tran (Rutgers University); River Bartz, Ram Durairajan (University of Oregon); Ulrich Kremer, Sudarsun Kannan (Rutgers university)

Emerald Tiers: Focusing on SSD+MAID Through a Green Lens
[Paper] [Slides]
Zhaokang Ke, Jim Diehl (University of Minnesota); Ya-Shu Chen (National Taiwan University of Science and Technology); David H.C. Du (University of Minnesota)


Session 6: Storage and Machine Learning

RingSampler: GNN Sampling on Large-Scale Graphs with io_uring
[Paper] [Slides]
Qixuan Chen, Yuhang Song, Melissa Martinez, Vasiliki Kalavri (Boston University)

Quantum Neural Networks Need Checkpointing
[Paper] [Slides]
Christopher Kverne, Mayur Akewar (Florida International University); Yuqian Huo, Tirthak Patel (Rice University); Janki Bhimani (Florida International University)

Can LLMs Model the Environmental Impact on SSD?
[Paper] [Slides]
Mayur Akewar, Gang Quan (Florida International University); Sandeep Madireddy (Argonne National Laboratory); Janki Bhimani (Florida International University)


Session 7: System Overheads and Security

Side-Channel Information Leakage with CPU Frequency Scaling (but without CPU Frequency)
[Paper] [Slides]
Li Zhu, Chundong Wang (ShanghaiTech University)

Bridging a Shortcut to Exempt the Software Tax Charged on Logging I/Os
[Paper] [Slides]
Yanpeng Hu, Yunxin Yang, Li Zhu, Chundong Wang (ShanghaiTech University)