At the 2026 Open AI Infra Summit, Alibaba Cloud showcased the evolution of the UALink 2.0 protocol and its integration with CXL, detailing new specifications, in‑network compute capabilities, and ecosystem developments that aim to overcome scale‑up bottlenecks in AI training and inference.
The article explains Google’s TPU data‑center architecture, detailing the vertical Scale‑Up strategy within a SuperPod, the horizontal Scale‑Out across SuperPods, the 3D Torus topology with Twisted variants, and the multi‑layer network design that enables petabyte‑scale AI training and inference.
As AI models grow to massive scales, the demand for ultra‑high‑performance, low‑latency networking in xPU clusters intensifies, prompting a shift from dense to MoE architectures and driving the evolution of Scale‑up networks, where Alibaba Cloud’s UPN design tackles bandwidth, cost, and reliability challenges.
This article introduces four cutting‑edge AI networking technologies—SUE, OISA, ALS, and ETH+—detailing their backgrounds, architectural designs, and performance enhancements that enable ultra‑high bandwidth, low‑latency, and scalable interconnects for modern AI compute clusters.
This report analyzes how the convergence of Moore's law slowdown and large‑model scaling laws creates a feedback loop between compute power and intelligence, driving the emergence of AI‑specific chips, high‑speed networking, and DPU architectures that together reshape modern AI infrastructure.
Large‑scale AI model training now demands unprecedented bandwidth and latency performance, forcing data centers to adopt three scaling approaches—Scale‑up, Scale‑out, and Scale‑Across—while leveraging optical I/O, CPO, and optical circuit switching to overcome power, distance, and bandwidth limits.
This article explains the technical definitions, core differences, and practical use cases of Scale‑Up and Scale‑Out networking in AI systems, highlighting how they impact latency, bandwidth, and cost, and illustrates their combined application through NVIDIA's NVL72 supernode case study.
This article explains the concepts of AI Pods and GPU clusters, compares vertical (scale‑up) and horizontal (scale‑out) expansion, describes XPU types, discusses internal and inter‑pod communication, and evaluates the benefits and drawbacks of each scaling approach along with relevant networking technologies.
At the CCF China Storage Conference 2024 in Guangzhou, Alibaba Cloud’s research team presented the ALS (ALink System) ecosystem, discussed Scale‑Up interconnect protocols, and demonstrated multi‑layer storage innovations for AI workloads, highlighting hardware‑software integration, high‑bandwidth networking, and future CXL/PIM research.
The article details NetEase Cloud Music’s high‑concurrency cache architecture—using lazy‑load, hole‑wrapped objects for penetration protection, placeholder values for missing data, horizontal and vertical scaling with consistent hashing, and asynchronous binlog‑driven invalidation—to achieve sub‑millisecond reads for a read‑heavy, write‑light music library.
This article explains SAP HANA as an in‑memory database platform, detailing its appliance and cloud deployments, Scale‑Up, Scale‑Out, and TDI solution types, typical OLTP/OLAP scenarios, certification, virtualization support, storage sizing, and performance benchmarks.
This article provides a comprehensive overview of SAP HANA, covering its role as an in‑memory database, deployment models (cloud, appliance, on‑premise), primary application scenarios, hardware certification, scale‑up versus scale‑out architectures, TDI integration, virtualization support, storage sizing, high‑availability options and node roles.
This article explains SAP HANA’s role as an in‑memory database platform, outlines its cloud and on‑premise deployment models, describes key business scenarios such as Business Warehouse on HANA and Business Suite on HANA, and details scale‑up vs. scale‑out, TDI, virtualization, storage sizing, and high‑availability configurations.
This article explains why high‑concurrency design is essential for modern internet services, compares vertical (scale‑up) and horizontal (scale‑out) scaling, revisits the C10K/C10M challenges, and reviews process, thread, coroutine, and event‑driven models to help engineers build ultra‑scalable web servers.
