China

Solution NIO adopted a hybrid GPU sharing strategy, selecting different GPU allocation mechanisms based on workload characteristics rather than enforcing a single approach. The following diagram illustrates the GPU management architecture after integrating HAMi into Kubernetes....

The Challenge: Resource fragmentation As Xiaohongshu’s business grew rapidly, its infrastructure demands quickly outpaced the capacity of individual Kubernetes clusters. Both self-built data centers and managed Kubernetes services (TKE/ACK) from cloud vendors have cluster size limits...

Challenge As Ke Holdings’ machine learning initiatives scaled, the infrastructure team faced significant challenges in GPU resource management: Initially, the overall GPU utilization was only 13% due to the complexity of the multi-cloud environment and diverse...

Solution DaoCloud adopted HAMi, a CNCF Sandbox project, for heterogeneous AI computing virtualization, as the unified GPU layer across both D.run and DCE. HAMi provides device virtualization, vGPU partitioning, and scheduling for heterogeneous accelerators in Kubernetes...

EffectiveGPU leverages several key capabilities provided by HAMi: EffectiveGPU further extends the capabilities of HAMi and provides the following advanced features: By leveraging these technologies, the team can deploy more services with fewer GPUs, dynamically adjust...

Implementing Cilium for Superior Cloud Native Networking QingCloud’s journey with Cilium began in 2019 when their team noticed the project and recognized its superior networking performance over similar CNI solutions. “In 2019, some of our customers...

Switching To Cilium For Scalable and Cloud Native Networking Trip.com Group Limited, a multinational travel service conglomerate, serves customers in over 40 languages and 200 countries. Their operations are supported by a vast IT infrastructure, with...

Oppo is China’s largest smartphone manufacturer, holding the fifth largest market share of smartphone manufacturers globally, with devices sold in more than 200,000 retail outlets. In 2021, AI pioneers Oppo built a large-scale, end-to-end machine learning...