Official service account of 360 Smart Cloud, dedicated to building a high-quality, secure, highly available, convenient, and stable one‑stop cloud service platform.
This article reviews distributed training techniques for large deep‑learning models, covering data parallelism, model parallelism (including pipeline and tensor parallelism), gradient bucketing and accumulation, 3D parallelism, and practical implementations such as Megatron‑LM and 360AI platform optimizations.
The article describes how a company implements an idle compute sharing feature for dedicated Kubernetes clusters, leveraging Karmada to allocate spare CPU and memory to offline workloads, thereby improving resource utilization, reducing costs, and outlining usage scenarios, configuration steps, technical architecture, and future plans.
This article presents a comprehensive overview of Ozone, a Hadoop‑based distributed object storage system, detailing its architecture, metadata management, scalability enhancements, small‑file handling, erasure coding, lifecycle policies, and future improvements aimed at boosting performance and reliability for large‑scale unstructured data workloads.
The article analyzes low CPU utilization in pure online Kubernetes clusters, introduces mixed‑workload (online + offline) scheduling to improve resource efficiency, explains core techniques, kernel QoS features, and details Koordinator‑based implementations such as node resource reservation and scheduling adjustments.
This article explains what hybrid cloud is, outlines its various forms, describes its architectural features, presents typical scenarios such as scaling compute, storage, security and product services, and discusses the challenges faced in hybrid cloud adoption.
The Pika community announced version 3.5.2, fixing long‑standing bugs and adding features such as Redis transaction support, a cache layer for hot‑cold data separation, dynamic Codis‑Proxy configuration, and performance improvements, while also outlining the planned enhancements for the upcoming 3.5.3 release.
This article explains how asynchronous programming and event‑driven mechanisms work in both frontend JavaScript and backend Golang, covering the JavaScript event loop, macro‑ and micro‑tasks, goroutine‑based concurrency, and the kernel‑level epoll architecture that together enable efficient, non‑blocking execution.
This article provides a comprehensive overview of WebRTC's MediaStream and RTCPeerConnection APIs, explaining core concepts such as tracks, sources, sinks, device enumeration, media constraints, bitrate and resolution settings, compatibility considerations, screen sharing, content hints, and the step‑by‑step workflow for establishing a peer‑to‑peer connection in the browser.
This article summarizes a collection of newly released open‑source AI projects covering natural‑language processing, multimodal processing, intelligent agents, recommendation systems, and model training acceleration, providing brief descriptions, key capabilities, and links to their repositories.
This article introduces Elasticsearch, explains its hot‑warm‑cold lifecycle management, demonstrates vector search and built‑in NLP capabilities, and describes how the 360 Zhihui Cloud Platform integrates these features with practical test cases and new visualization tools.
