Tagged articles
AI workloads
23 articles · Page 1 of 1
Running AI workloads on Kubernetes introduces five production‑grade challenges—complex GPU and accelerator management, workload‑aware scheduling, inference autoscaling beyond CPU metrics, multi‑layer observability, and Day 2 governance—requiring platform teams to extend their capabilities beyond traditional container operations.
AI workloadsDay 2 operationsGPU scheduling
0 likes · 10 min read

ACK Pro introduces a provisioned control‑plane mode that replaces reactive scaling with preset performance tiers, guaranteeing deterministic capacity for thousands of nodes and tens of thousands of Pods, and a real‑world AI training case shows reduced pod‑startup latency, eliminated HTTP 429 errors, and about 30% faster training cycles.
ACK ProAI workloadsKubernetes
0 likes · 10 min read

Brendan Burns recounts how he convinced Google to back Kubernetes, built the MVP in five days, navigated open‑source governance, tackled technical challenges like Etcd and declarative design, expanded the platform for AI workloads, and reflects on why even successful software like Kubernetes inevitably faces obsolescence.
AI workloadsKubernetescloud native
0 likes · 31 min read

The article presents a detailed June 2026 GPU performance ranking, dividing graphics cards into five tiers—from flagship 300X‑200X models like RTX 5090 and RX 7900 XTX for 4K/8K gaming and AI workloads, down to legacy integrated GPUs for basic office tasks—while also summarizing the strengths of Apple/Intel, NVIDIA, AMD, and APU solutions.
AI workloadsAMD RadeonGPU
0 likes · 6 min read

The Dawn 8000 (Dengfeng) AI supercluster, China’s first fully domestic 100‑k‑card system, overcame long‑standing compute bottlenecks by integrating native super‑intelligent fusion, six self‑developed chips, a lossless RDMA network, high‑density liquid‑cooled cabinets, and a unified scheduling platform, achieving full‑load operation in its first week at WAIC.
AI supercomputingAI workloadsRDMA network
0 likes · 14 min read

China’s Sugon ParaStor F9000 all‑flash distributed storage system has claimed the top spots on both the full‑node and 10‑node IO500 production benchmarks, delivering up to 247% bandwidth gains and powering large‑scale AI, scientific simulation, and autonomous‑driving workloads across a ten‑thousand‑card supercomputing cluster.
AI workloadsAll-flash storageChinese hardware
0 likes · 9 min read

Kubernetes 1.36 marks a shift toward native governance of complex AI resources, workload‑aware scheduling, fine‑grained security, and control‑plane scalability, urging platform teams to rethink resource allocation, isolation, and governance as AI workloads move into production.
AI workloadsDRAKubernetes
0 likes · 10 min read

The article analyzes the evolving server CPU landscape, forecasting a three‑way split among Intel, AMD, and ARM by 2030, driven by AI workloads, shifting CPU‑GPU ratios, supply constraints, price hikes, and rapid revenue growth exceeding $200 billion.
AI workloadsAMDARM
0 likes · 8 min read

The article explains how traditional Parquet‑based lakehouse architectures, optimized for large‑scale scans, struggle with AI workloads that need ultra‑low‑latency random access, and how Lance redesigns the storage format, indexing and write path to provide O(1) addressing, native vector support, and seamless integration with native execution engines.
AI workloadsLanceNative Execution
0 likes · 12 min read

This article analyzes Milvus—its open‑source origins, three deployment modes, four‑layer architecture, eight‑plus indexing algorithms, real‑world case studies, and a detailed comparison with competitors—highlighting its strengths, weaknesses, common pitfalls, and when it’s the right choice for large‑scale AI workloads.
AI workloadsMilvusVector Database
0 likes · 15 min read


21CTOFeb 2, 2026 · Databases
Oracle claims a new era for MySQL by moving commercial‑only features to the community edition and adding AI‑focused vector functions, but developers question whether these promises are timely or sufficient, fearing continued neglect of the open‑source community.
AI workloadsDatabase CommunityMySQL
0 likes · 5 min read

The article analyzes NVIDIA's newly announced cuTile, a tile‑based Python DSL for GPU kernels, examining its technical differences from CUDA's SIMT model, its potential to reshape the GPU programming ecosystem, community reactions, competition with Triton, and the uncertain future that hinges on ecosystem maturity and migration tools.
AI workloadsCUDAGPU programming
0 likes · 12 min read

Explore the 15 most impactful Kubernetes scenarios—from microservices and auto‑scaling to multi‑cloud deployments, AI workloads, edge computing, and compliance—detailing how they boost reliability, efficiency, and cost‑effectiveness, while also highlighting situations where Kubernetes may not be the right choice.
AI workloadsAuto ScalingEdge Computing
0 likes · 11 min read

Fluid introduces a cloud‑native elastic data abstraction that lets AI workloads efficiently access, manage, and accelerate heterogeneous data sources across serverful and serverless environments, offering unified Dataset, Runtime, and DataOperation concepts, and has been recognized by CNCF’s 2024 Technology Radar.
AI workloadsCNCFFluid
0 likes · 9 min read

This article presents a detailed comparison of major NVIDIA GPUs—including RTX 4090, RTX 4090 D, RTX 3090, A10, A40, A100, and H100—covering memory size, bandwidth, Tensor BF16/FP16/FP32 throughput, FP16/FP32 performance, power draw and release dates, and explains how these specs affect AI workload efficiency.
AI workloadsGPUNVIDIA
0 likes · 9 min read

The article analyzes how Co‑Packaged Optics (CPO) and silicon photonics address exploding data‑center bandwidth demands, reduce power consumption, and enable AI‑driven workloads, while outlining industry roadmaps, major vendor contributions, and future technical challenges.
AI workloadsTSMC COUPEco-packaged optics
0 likes · 14 min read

In an interview, Alibaba Cloud’s distributed cloud‑native platform lead explains how distributed cloud‑native addresses elasticity, high‑availability, and multi‑cluster management challenges, outlines the evolution of ACK One, and forecasts its role in AI and edge computing for modern enterprises.
ACK OneAI workloadsEdge Computing
0 likes · 11 min read

The article explains the mGPU solution from Volcano Engine, detailing its kernel‑level GPU virtualization, container runtime hooks, and scheduling mechanisms that allow multiple containers to share a single NVIDIA GPU with isolated compute and memory resources, achieving near‑lossless performance and up to 50% higher utilization for AI tasks.
AI workloadsGPU Sharingcontainer-runtime
0 likes · 9 min read

The article examines the rapid growth of AI‑driven compute demand, explains why conventional ECMP load balancing struggles with uneven, high‑bandwidth flows in data‑center networks, and compares advanced strategies such as Fat‑Tree design, VoQ, flow‑based, packet‑based, flowlet, and cell‑based approaches, including vendor implementations.
AI workloadsData Center NetworkECMP
0 likes · 13 min read

As AI workloads push intelligent compute power growth beyond 50% CAGR, data‑center networks face massive parallel paths, making traditional ECMP load‑balancing insufficient and causing severe congestion, while newer granular schemes such as packet‑spraying, flowlet, and cell‑based balancing offer higher bandwidth utilization and fairness.
AI workloadsECMPRPS
0 likes · 17 min read

This article explains how Kubernetes supports GPU workloads for AI and machine learning, covering device plugins, pod GPU requests, oversubscription, security isolation, cloud‑provider node setup, and protecting GPU nodes from non‑GPU pods.
AI workloadsDevice PluginGPU
0 likes · 8 min read

Baidu Intelligent Cloud’s GPU Container Virtualization 2.0 combines user‑mode and kernel‑mode isolation in a dual‑engine design that unifies scheduling of AI compute, rendering and encoding, supports mixed deployment and multi‑scheduler integration, and boosts GPU utilization across inference, offline tasks, autonomous‑driving simulation, and cloud‑gaming workloads.
AI workloadsGPU virtualizationRendering isolation
0 likes · 14 min read

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.
AI workloadsGPU virtualizationKubernetes scheduling
0 likes · 14 min read
