Empowering Azure Storage with RDMA
Wei Bai, Shanim Sainul Abdeen, Ankit Agrawal, Krishan Kumar Attre, Paramvir Bahl, Ameya Bhagat, Gowri Bhaskara, Tanya Brokhman, Lei Cao, Ahmad Cheema, Rebecca Chow, Jeff Cohen
Abstract
Given the wide adoption of disaggregated storage in public clouds, networking is the key to enabling high performance and high reliability in a cloud storage service. In Azure, we choose Remote Direct Memory Access (RDMA) as our transport and aim to enable it for both storage frontend traffic (between compute virtual machines and storage clusters) and backend traffic (within a storage cluster) to fully realize its benefits. As compute and storage clusters may be located in different datacenters within an Azure region, we need to support RDMA at regional scale.
This work presents our experience in deploying intra-region RDMA to support storage workloads in Azure. The high complexity and heterogeneity of our infrastructure bring a series of new challenges, such as the problem of interoperability between different types of RDMA network interface cards. We have made several changes to our network infrastructure to address these challenges. Today, around 70% of traffic in Azure is RDMA and intra-region RDMA is supported in all Azure public regions. RDMA helps us achieve significant disk I/O performance improvements and CPU core savings.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext eb62fc03-3825-4656-9343-baed4353ddbeCited by top-tier papers51
- Alibaba HPN: A Data Center Network for Large Language Model TrainingKun Qian, Yongqing Xi, Jiamin Cao, Jiaqi Gao et al.SIGCOMM 2024 · 173 citations
- ServerlessLLM: Low-Latency Serverless Inference for Large Language ModelsYao Fu, Leyang Xue, Yeqi Huang, Andrei-Octavian Brabete et al.OSDI 2024 · 125 citations
- Towards Domain-Specific Network Transport for Distributed DNN TrainingHao Wang, Han Tian, Jingrong Chen, Xinchen Wan et al.NSDI 2024 · 54 citations
- Reverie: Low Pass Filter-Based Switch Buffer Sharing for Datacenters with RDMA and TCP TrafficVamsi Addanki, Wei Bai, Stefan Schmid, Maria ApostolakiNSDI 2024 · 33 citations
- White-Boxing RDMA with Packet-Granular Software ControlChenxingyu Zhao, Jaehong Min, Ming Liu, Arvind KrishnamurthyNSDI 2025 · 28 citations
Builds on25
- A Unified Architecture for Accelerating Distributed DNN Training in Heterogeneous GPU/CPU ClustersYimin Jiang, Yibo Zhu, Chang Lan, Bairen Yi et al.OSDI 2020 · 390 citations
- Swift: Delay is Simple and Effective for Congestion Control in the DatacenterGautam Kumar, Nandita Dukkipati, Keon Jang, Hassan M. G. Wassel et al.SIGCOMM 2020 · 333 citations
- Jupiter evolving: transforming google's datacenter network via optical circuit switches and software-defined networkingLeon Poutievski, Omid Mashayekhi, Joon Ong, Arjun Singh et al.SIGCOMM 2022 · 230 citations
- When Cloud Storage Meets RDMAYixiao Gao, Qiang Li, Lingbo Tang, Yongqing Xi et al.NSDI 2021 · 228 citations
- SRNIC: A Scalable Architecture for RDMA NICsZilong Wang, Layong Luo, Qingsong Ning, Chaoliang Zeng et al.NSDI 2023 · 154 citations
Related papers
- CoRM: Compactable Remote Memory over RDMAKonstantin Taranov, Salvatore Di Girolamo, Torsten HoeflerSIGMOD 2021 · 18 citations
- SF-STACK: Streamlining RDMA for Heterogeneous Telecom StorageJian Tang, Wenming Zheng, Xiaoliang Wang, Xiaoping Fan et al.INFOCOM 2026
- FineMem: Breaking the Allocation Overhead vs. Memory Waste Dilemma in Fine-Grained Disaggregated Memory ManagementXiaoyang Wang, Yongkun Li, Kan Wu, Wenzhe Zhu et al.OSDI 2025 · 3 citations
- SmartDS: Middle-Tier-centric SmartNIC Enabling Application-aware Message Split for Disaggregated Block StorageJie Zhang, Hongjing Huang, Lingjun Zhu, Shu Ma et al.ISCA 2023 · 15 citations
- Outback: Fast and Communication-efficient Index for Key-Value Store on Disaggregated MemoryYi Liu, Minghao Xie, Shouqian Shi, Yuanchao Xu et al.VLDB 2025 · 5 citations
