This article explores Meta's Llama 2 large language model—its performance, expanded training data, architectural details, evaluation results, RLHF fine‑tuning process, and step‑by‑step deployment on UCloud UK8S using Docker and Kubernetes—providing a comprehensive guide for AI practitioners.
This article reviews the architectures, training methods, and key characteristics of major open‑source large language models such as BERT, GPT, T5, LLaMA and ChatGLM, and then provides a step‑by‑step tutorial for deploying ChatGLM2‑6B on UCloud's UK8S platform using UFS storage, Kubernetes manifests, and command‑line tools.
This article walks you through the complete installation and configuration of UCloud's free USDP (UCloud Data Platform) on a three‑node CentOS 7.2‑7.6 cluster, covering environment preparation, package download, repair scripts, MySQL setup, service startup, web UI activation, monitoring, and a quick Hive query example.
This article provides a comprehensive tutorial on installing UCloud's free USDP version for private big‑data deployments, covering environment preparation, minimum node specifications, resource download, configuration files, one‑click initialization scripts, server startup, web UI access, license acquisition, and optional manual setup procedures.
This article explains why many enterprises still rely on on‑premise data centers, introduces three hybrid‑cloud deployment models, and provides detailed solutions for extending computing power, storage backup, security protection, new product capabilities, and smooth business migration using a hybrid cloud approach.
This article details how UCloud’s USnap service extends the UDataArk CDP technology to provide low‑cost, second‑level snapshots for ordinary, SSD, and RSSD cloud disks, describing the client‑side IO capture, Stream‑based front layer redesign, Arker storage architecture, and efficient rollback mechanisms.
Facing outdated VMware servers and costly data center upkeep, a restaurant client migrated its workloads to UCloud using a hybrid approach of public‑bandwidth transfer and offline disk copying, leveraging KVM, virt‑v2v, NBD, and custom APIs to ensure minimal downtime, cost savings, and scalable future growth.
This guide walks through setting up a quick image porn detection platform using UCloud's UAI‑Censor service, covering account creation, obtaining API keys, testing the API, and integrating a simple Java SpringBoot client that evaluates images and returns pass, forbid, or manual‑review suggestions.
In a detailed interview, UCloud’s virtual network lead Zhou Jian explains how the company’s SDN‑based solutions address heterogeneous network challenges—security isolation, performance, and consistent user experience—through programmable VPCs, hybrid gateways, global traffic cleaning, and multi‑cloud connectivity, illustrating the evolution from VPC 1.0 to today’s global backbone.
The article details the background, user pain points, and technical design of UCloud’s Cube serverless container product, explaining its firecracker‑based runtime, cri‑o integration, Kubernetes scheduling, storage and network solutions, and future optimization plans.
This article explains how UCloud reduced its data center Power Usage Effectiveness to around 1.3 through advanced cooling technologies, energy‑efficient UPS, modeling, and other green measures, while also reviewing international benchmarks and China’s regulatory targets for sustainable data centers.
A Chinese game studio faced a massive 300 GB DDoS ransom attack, refused to pay, and, with UCloud's elastic high‑availability and Anycast cleaning technologies, repelled the assault while detailing the attackers' methods and offering a public‑cloud DDoS mitigation guide.
This lesson explains the value and implementation of high‑availability systems, explores load‑balancing concepts, features, and scenarios—including health checks, session persistence, and scheduling algorithms—and introduces virtual IP solutions for robust cloud infrastructure.
This article explains how UCloud’s internal VIP solution provides second‑level failover for high‑availability services in cloud overlay networks, detailing traditional approaches, the evolution of broadcast mechanisms, and real‑world use cases such as e‑commerce payment systems and database services.
This article explains how UCloud's Rome product leverages HashiCorp Packer, combined with GitLab, Jenkins, and Terraform, to automate multi‑cloud image creation, dramatically reducing build time from hours to minutes while improving reliability and consistency across dozens of data centers.
This article explains how UCloud's internal‑VIP based high‑availability solution provides second‑level failover for cloud and physical hosts, detailing the evolution from simulated broadcast to a GARP‑sniffing broadcast cluster and showcasing real‑world e‑commerce and database use cases.
This article defines cloud computing as a distributed‑computing model that splits large tasks into many small programs processed across multiple servers, outlines its four key domains—computing, networking, storage, and applications—and highlights UCloud’s recent free‑course event and QR‑code registration for students and professionals.
The article explains how the WeiChe app leverages AI to instantly recognize car dashboard warning lights, describes the underlying deep‑learning infrastructure on UCloud’s UAI‑Inference platform, and shows how the Hot‑Standby feature dramatically cuts GPU costs while maintaining real‑time performance.
This article explains the evolution of DDoS protection from early router‑based defenses to modern cloud‑native solutions, detailing UCloud's three‑module architecture, the role of Anycast in traffic dilution, and practical features that enable scalable, cost‑effective mitigation of large‑scale attacks.
This article explains how UCloud's Packer builder automates the creation of consistent, versioned cloud host images, compares it with manual console methods, outlines the full lifecycle, provides code examples, and shows how combining Packer with Terraform enables powerful multi‑cloud Infrastructure as Code workflows.
The article recaps a Shanghai UCan afternoon tea tech salon where six seasoned experts shared practical Kubernetes and cloud‑native experiences, covering UCloud’s UK8S platform, network and storage integrations, tenant isolation, deployment automation, real‑world case studies from healthcare, workflow engines, service mesh, and small‑team adoption strategies.
This article explains how UCloud’s Packer builder automates the creation of consistent, versioned machine images across multiple cloud platforms, compares it with manual console methods, outlines the full lifecycle, provides practical JSON examples, and demonstrates integration with Terraform for seamless DevOps workflows.
UCloud’s engineering team built a Terraform‑based resource orchestration platform that cuts provisioning time from over three minutes to under a minute, enhances reliability through custom diffs, idempotent retries, and CI/CD integration, and supports multi‑cloud management, showcasing best practices in cloud‑native infrastructure automation.
This article explains the risks of using Terraform without proper safeguards, demonstrates a real-world failure caused by careless use of the --auto-approve flag, and presents a robust blue‑green deployment pattern with an arbitration module that enforces safe state transitions on UCloud.
This article explains how UCloud built a Terraform‑based resource orchestration tool that dramatically reduces deployment time, improves reliability, and integrates with CLI and CI/CD pipelines, while detailing fault‑tolerance, integration, and engineering practices for large‑scale cloud infrastructure.
This article details UCloud's five‑month, multi‑department hot migration of its Seoul data center, covering planning, ZooKeeper scaling, udatabase and MySQL migration strategies, deployment platforms, and the final cut‑over steps that ensured zero user impact.
This article explains how UCloud addresses IPv4 limitations by deploying a free IPv6 conversion service built on stateful NAT64 and programmable P4 switches, detailing the strategic rollout, system architecture, high‑availability and security mechanisms, load‑balancing algorithms, P4 table optimizations, and performance test results.
UCloud’s physical network orchestrator automates large‑scale data‑center switch configuration, reducing IDC network build cycles from 2‑3 days to 2‑3 hours, boosting success rates to 99%, while handling 3000+ switches, 200 Gbps access throughput, and supporting hybrid‑cloud real‑time connectivity through a scenario‑driven, Kafka‑backed architecture.
This article explains why UCloud introduced a Service Mesh, how Istio was chosen and adapted for the UAEK Kubernetes platform—including IPv6 challenges, performance testing, and future improvements—providing practical insights for building reliable micro‑service infrastructures.
Edge computing distributes compute and storage to network edge nodes, reducing latency and bandwidth usage, while UCloud's dynamic acceleration network leverages optimized routing, private UTP protocol, and seamless integration with edge platforms to enhance data transmission across diverse, often weak, network environments.
The 2018 UCloud Think in Cloud conference showcased how cloud infrastructure accelerates AI applications in customer service, education, and embedded devices, while also presenting blockchain use cases and security challenges across finance, insurance, and IoT, highlighting practical solutions and emerging platforms.
This article explains the design, components, and implementation of UCloud's application hot‑patch framework, covering its motivation, safety checks, multi‑thread support, and how the Creator, Loader, and Core Runtime work together to apply, manage, and roll back patches without restarting services.
In an exclusive interview, UCloud’s R&D Director Ye Lideng shares how the company’s CBA strategy, AI‑first products, and innovative cloud services empower startups, drive industry adoption, and shape the future of artificial intelligence and cloud computing in China.
This article details UCloud's comprehensive optimizations for KVM live migration—including host selection, zero‑data handling, network switch improvements, auto‑converge tuning, compression, fast‑migration paths, cross‑machine type migration, and local hot‑upgrade—demonstrating how each stage reduces downtime, data transfer, and resource consumption while maintaining VM availability.
This article explains the principles and detailed three‑stage process—preparation, migration, and switch—of KVM‑based online virtual machine migration on UCloud’s platform, illustrating each step with diagrams and highlighting challenges, benefits, and current limitations of non‑shared storage migration.
UCloud’s Compute Factory enables rapid provisioning of massive VM resources for compute‑intensive services by leveraging a Mesos‑based resource management platform that unifies multi‑region data centers, supports both VMs and containers, and addresses challenges in scheduling, networking, storage, and operational reliability.
The CNUTCon Global Operations Conference highlighted UCloud's container and Serverless practices, explaining the evolution from virtualization waste to function‑as‑a‑service, showcasing UCloud's universal compute platform and real‑world use cases such as image processing, gene data analysis, and OCR recognition.
The article explains why thousands of MongoDB instances are repeatedly compromised—due to password‑less logins and public exposure—details the inherent design flaws, and describes UCloud’s UDB MongoDB security measures such as mandatory authentication, VPC isolation, data backup, and performance‑friendly connection pooling.
This guide walks you through installing StrongSwan on a CentOS client, configuring IPSec tunnels, and creating a UCloud IPSec VPN gateway to achieve secure hybrid‑cloud network connectivity using VPN.
This article explains the Serverless concept, compares it with PaaS and BaaS, and provides a step‑by‑step guide to building an AI‑driven Van Gogh style‑transfer service using UCloud's Serverless (UGC) platform, covering both backend setup and frontend interaction.
This guide walks you through training an Inception‑V3 model on the FER2013 dataset with TensorFlow 1.1, packaging the model, and deploying a scalable facial expression recognition API using UCloud's AI‑as‑a‑Service platform, including performance testing against GPU benchmarks.
