XConn Technologies, in Collaboration with Alibaba Cloud, Wins Prestigious ACM SIGMOD 2025 Industry Track Best Paper Award for Breakthrough Work on CXL-Based Disaggregated Memory
Award-winning research highlights the potential of CXL memory pooling in commercial cloud-native database deployments, setting the stage for next-generation AI and HPC workloads
SANTA CLARA, Calif. – July 8, 2025 – XConn Technologies (XConn), the innovation leader in next-generation interconnect technology for the future of high-performance computing and AI applications, today announced that a whitepaper XConn co-authored with Alibaba Cloud has been awarded the prestigious Industry Track Best Paper Award at ACM SIGMOD 2025, the premier international conference on data management.
​
The paper, titled "Unlocking the Potential of CXL for Disaggregated Memory in Cloud-Native Databases," presents the first scalable implementation of CXL (Compute Express Link®) memory pooling and sharing for in-memory databases in real-world cloud environments. The award recognizes the technical depth and commercial relevance of the collaboration between XConn and Alibaba Cloud.
​
The paper introduces PolarCXLMem, a CXL switch-based disaggregated memory system built using XConn’s leading-edge CXL switch technology, and demonstrates its integration with PolarDB, a widely deployed cloud-native database from Alibaba Cloud, the digital technology and intelligence backbone of Alibaba Group. Through innovative designs including the PolarRecv instant recovery scheme and a novel cache coherency protocol, the system enables seamless memory pooling, instant crash recovery, and efficient multi-node data sharing.
​
"This recognition from ACM SIGMOD is a major milestone for the industry and a proud moment for our team," said Gerry Fan, CEO of XConn Technologies. "For the past three years, we’ve worked closely with Alibaba Cloud to validate the real-world potential of CXL in data-intensive cloud environments. This paper proves that CXL-based memory pooling is not just theoretical—it’s scalable, commercially viable, and a foundational enabler for the future of AI and HPC."
​
“Our collaboration with XConn shows how CXL can transform cloud-native infrastructure by enabling efficient and resilient memory disaggregation,” said Feifei Li, Vice President of Alibaba Cloud Intelligence Group, Head of Database Products, Alibaba Cloud Intelligence. “Through this work, we are redefining cloud native database’s performance, scalability, and fault tolerance for cloud applications at scale.”
​
In performance evaluations under a variety of workloads, PolarCXLMem demonstrated up to 2.1× throughput improvement in memory pooling scenarios and 1.55× improvement in sharing scenarios compared to RDMA-based systems. These results mark a significant step forward in using CXL for high-performance, cost-efficient cloud database architectures.
​
For more information the full paper is available for download here: https://lnkd.in/gwwB_4Ph
About XConn Technologies
XConn Technologies Holdings, Inc. (XConn) is the innovation leader in next-generation interconnect technology for high-performance computing and AI applications. The company is the industry’s first to deliver a hybrid switch supporting both CXL 2.0 and PCIe Gen5 on a single chip. Privately funded, XConn is setting the benchmark for data center interconnect with scalability, flexibility and performance. For more information visit: xconn-tech.com.
​
Media Contacts:
Steve Sturgeon
Avista PR
858.472.5669