venue: SoCC
year: 2025
papers:
- title: 'From Bottleneck to Breakthrough: Optimizing Scheduling for Hyperscale Containerized
    Clusters'
  authors:
  - Bing Li
  - Yuquan Ren
  - Xinyi Song
  - Zhilei Liu
  - Cong Xu
  - Jingyuan Zhang
  - Caixue Lin
  - Wu Xiang
  - Rui Shi
  reason: "Documents production-scale scheduling improvements at a hyperscale cloud provider, demonstrating how targeted optimizations reduce scheduling tail latency and increase cluster utilization in real containerized workloads."
- title: 'CPU-Limits kill Performance: Time to rethink Resource Control'
  authors:
  - Chirag C. Shetty
  - Sarthak Chakraborty
  - Hubertus Franke
  - Larisa Shwartz
  - Chandra Narayanaswami
  - Indranil Gupta
  - Saurabh Jha
  reason: "Challenges the conventional use of CPU cgroup limits in cloud environments, showing through production evidence that CFS bandwidth throttling degrades application QoS and proposing a rethink of resource control abstractions."
- title: Rethinking Tiered Memory Management in Cloud Data Centers
  authors:
  - Tong Xing 0002
  - Jiaxun Yang
  - Javier Picorel
  - Antonio Barbalace
  reason: "Proposes a novel tiered memory management framework for cloud data centers that improves performance by rethinking the placement and migration policies across DRAM and CXL/NVM tiers."
- title: Cost-Efficient Cloud Infrastructure with Hugepage-aware Memory Deduplication
  authors:
  - Ruizhe Huang
  - Xinyu Wang 0043
  - Zhida An
  - Hanwen Lei
  - Peng Jiang 0007
  - Ziqi Zhang
  - Ding Li 0001
  - Yao Guo 0001
  - Xiangqun Chen
  - Yuntao Liu
  - Kang Zhou
  - Yuxin Ren 0001
  - Ning Jia 0004
  - Xinwei Hu
  reason: "Deploys hugepage-aware memory deduplication in a large production cloud, achieving significant memory savings without the performance regressions that plague conventional THP-based deduplication."
- title: 'ALAP: Intent-Based Serverless Computing via Delayed Decision-Making'
  authors:
  - Prasoon Sinha
  - Kostis Kaffes
  - Neeraja J. Yadwadkar
  reason: "Introduces an intent-based programming model for serverless that defers scheduling decisions until runtime context is available, improving resource efficiency and SLO attainment over eager placement strategies."
- title: 'Hydra: Virtualized Multi-Language Runtime for High-Density Serverless Platforms'
  authors:
  - Serhii Ivanenko
  - Vasyl Lanko
  - Rudi Horn
  - Vojin Jovanovic
  - Rodrigo Bruno
  reason: "Presents a virtualized runtime that multiplexes multiple language environments within a single sandbox, enabling higher function density and faster cold starts on serverless platforms."
- title: 'Serverless Elasticsearch: the Architecture Transformation from Stateful to
    Stateless'
  authors:
  - Iraklis Psaroudakis
  - Pooya Salehi
  - Jason Bryan
  - Francisco Fernández Castaño
  - Brendan Cully
  - Ankita Kumar
  - Henning Andersen
  - Thomas Repantis
  reason: "Describes Elastic's production migration of Elasticsearch to a serverless, stateless architecture, sharing engineering lessons on decoupling compute from state at cloud scale."
- title: 'DFUSE: Strongly Consistent Write-Back Kernel Caching for Distributed Userspace
    File Systems'
  authors:
  - Haoyu Li
  - Jingkai Fu
  - Qing Li 0002
  - Windsor Hsu
  - Asaf Cidon
  reason: "Closes a long-standing gap in FUSE-based distributed file systems by enabling strongly consistent write-back caching in the kernel, significantly improving throughput without sacrificing correctness."
- title: Accelerating Distributed Filesystem Metadata Service via Decoupling Directory
    Semantics from Metadata Indexing
  authors:
  - Wenhao Lv
  - Hao Guo
  - Qing Wang 0031
  - Youyou Lu
  - Jiwu Shu
  reason: "Achieves scalable distributed filesystem metadata by separating directory namespace semantics from the underlying index structure, reducing contention and improving throughput for large-scale cloud storage."
- title: 'Valet: Efficient Data Placement on Modern SSDs'
  authors:
  - Devashish R. Purandare
  - Peter Alvaro
  - Avani Wildani
  - Darrell D. E. Long
  - Ethan L. Miller
  reason: "Exploits fine-grained internal SSD geometry to make smarter data placement decisions, yielding measurable I/O performance gains without changes to the host storage stack."
- title: 'Understanding Diffusion Model Serving in Production: A Top-Down Analysis
    of Workload, Scheduling, and Resource Efficiency'
  authors:
  - Yanying Lin
  - Shuaipeng Wu
  - Shutian Luo
  - Hong Xu 0001
  - Haiying Shen
  - Chong Ma
  - Min Shen
  - Le Chen
  - Chengzhong Xu 0001
  - Lin Qu
  - Kejiang Ye
  reason: "Provides the first comprehensive production characterization of diffusion model inference workloads, revealing unique scheduling and resource efficiency challenges distinct from LLM serving."
- title: 'ModServe: Modality- and Stage-Aware Resource Disaggregation for Scalable
    Multimodal Model Serving'
  authors:
  - Haoran Qiu
  - Anish Biswas
  - Zihan Zhao
  - Jayashree Mohan
  - Alind Khare
  - Esha Choukse
  - Íñigo Goiri
  - Zeyu Zhang 0005
  - Haiying Shen
  - Chetan Bansal
  - Ramachandran Ramjee
  - Rodrigo Fonseca
  reason: "Disaggregates compute resources per modality and pipeline stage for multimodal inference, with a Microsoft production deployment showing improved GPU utilization and latency over monolithic serving."
- title: 'THORN-ML: Transparent Hardware Offloaded Resilient Networks for RDMA based
    Distributed ML Workloads'
  authors:
  - Maziyar Nazari
  - Daniel Noland
  - Giulio Sidoretti
  - Erika Hunhoff
  - Tamara Silbergleit Lehman
  - Eric Keller
  reason: "Offloads RDMA fault detection and recovery to programmable network hardware, making distributed ML training resilient to network failures without modifying the training framework or incurring software overhead."