We share the cloud tech topics you care about. Feel free to leave a message and tell us what you'd like to learn.
This article explains the fundamentals of vector retrieval, its role in enhancing large language models through embedding and prompt engineering, and details the algorithms, system architecture, and Baidu's engineering practices for building high‑performance vector databases.
This article examines the storage challenges posed by large AI models, analyzes the full workflow from data ingestion to inference, compares HDFS and object‑storage data lakes, and presents Baidu's cloud‑native storage‑acceleration solutions—including RapidFS and PFS—that dramatically improve training speed, checkpoint handling, and model deployment throughput.
This article explains Baidu Intelligent Cloud’s GPU container virtualization 2.0, detailing its dual‑engine architecture, resource pooling, and scheduling innovations that isolate AI, rendering, and codec workloads, and showcases real‑world scenarios such as online inference, autonomous‑driving simulation, and cloud gaming to improve GPU utilization.
This article examines the performance challenges of cloud‑native data lakes built on flat object storage and explains how a hierarchical‑namespace design improves directory operations, reduces request amplification, and delivers significant speedups for big‑data and AI workloads.
This article explains the design and engineering of Baidu's AIPod high‑performance network, detailing the massive bandwidth, scalability, stability, and low‑latency requirements of large‑scale AI model training and the practical tools used to monitor and troubleshoot such workloads.
This article explains the design and evolution of Baidu Cloud File System's (CFS) metadata service, detailing how a novel lock‑free architecture and strategic data layout enable POSIX‑compatible, highly scalable storage that can handle billions of files while maintaining high performance and consistency.
Baidu Intelligent Cloud’s 2023 GTC presentation details how its DPU‑based IaaS architecture unifies high‑performance compute, networking, storage, and security, addressing rapid AI workload growth, reducing CPU bottlenecks, and delivering elastic, cost‑effective solutions across virtual machines, bare‑metal servers, and specialized RDMA instances.
This article explains the limitations of traditional X86‑DPDK load‑balancing gateways and how Baidu’s third‑generation Universal Networking Platform (UNP) combines programmable ASICs, CPUs, and FPGA acceleration to deliver multi‑terabit throughput, ultra‑low latency, and dramatically lower cost and power consumption.
This article explains how Baidu built a massive, high‑performance GPU/IB cluster, optimized its architecture and software stack, and integrated AI frameworks and resource management to overcome compute, memory, and communication bottlenecks, enabling efficient training of trillion‑parameter large models.
The EuroSys 2023 paper on Baidu’s cloud‑native file storage CFS reveals a metadata service design that trims critical sections, achieving 1.76‑75.82× higher throughput and up to 91.71% lower latency compared to HopsFS and InfiniFS, and has been production‑stable for over three years.
