UCloud is a leading neutral cloud provider in China, developing its own IaaS, PaaS, AI service platform, and big data exchange platform, and delivering comprehensive industry solutions for public, private, hybrid, and dedicated clouds.
This article details the discovery, analysis, and resolution of a DPDK bug that caused a zero‑byte IP fragment to fill transmission queues, leading to packet loss and tx‑hang in UCloud's high‑availability ULB4 load balancer, and shares the tools and lessons learned for future debugging.
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.
The article uses the historic siege of Alesia to illustrate how Roman emphasis on infrastructure parallels modern DevOps practices, arguing that rapid, reliable delivery requires robust, automated cloud and public‑cloud foundations, while critiquing the myth of speed through long work hours and highlighting Netflix’s chaos engineering as a model for resilient systems.
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 how UCloud’s self‑developed IO acceleration technology dramatically narrows the CPU‑disk performance gap by raising 4K random write throughput from 300 IOPS to 45,000 IOPS, detailing its architecture, first‑generation limitations, and the second‑generation enhancements that enable scalable, fault‑tolerant cloud storage.
This article examines the rapid adoption of 25G SmartNICs for cloud workloads, compares major vendor solutions, explains why an Open vSwitch‑based TC Flower offload design was chosen, and details the practical challenges and performance gains achieved when deploying the solution in a public‑cloud environment.
UCloud’s latest SSD cloud disk redesign dramatically improves performance—raising IOPS by 13‑fold, cutting latency tenfold, and expanding capacity—through a two‑layer IO path, 1 MB metadata shards, multithreaded models, overload protection, online migration, and upcoming RDMA/SPDK‑based ultra‑high‑performance storage solutions.
The article investigates a Go 1.8.3 issue where goroutines block on ForkLock during fork‑exec, analyzes the kernel and Go runtime behavior, presents a hypothesis about memory‑heavy processes, validates it with experiments, and shows that upgrading to Go 1.9 or later eliminates the problem.
UCloud’s first Tech Talk detailed a comprehensive redesign of its cloud disk architecture, introduced ServiceMesh‑based gray‑scale deployment for virtual networking, and unveiled a second‑level snapshot system that delivers second‑level continuous backups with rapid restore capabilities, all backed by NVMe and programmable switch innovations.
