1 views collected around this technical thread.
This article explains how Douyin, TikTok, Baidu, Alibaba Cloud and Tencent operate self‑built data centers with terabit‑level outbound bandwidth, details ByteDance's server count growth from tens of thousands to hundreds of thousands, and describes the CDN technologies that enable billions of users to stream smoothly.
NVLink Fusion, unveiled at Computex 2025, extends NVIDIA’s NVLink technology to enable high‑bandwidth, low‑latency connections between CPUs and GPUs or third‑party accelerators, offering up to 900 GB/s bandwidth, flexible heterogeneous configurations, ecosystem expansion, performance gains for AI training and inference, and potential cost reductions.
This article explains how Douyin (TikTok) and other major Chinese platforms achieve massive concurrent usage by operating data centers with hundreds of thousands of servers, employing terabit-level outbound bandwidth, dual‑link designs, CDN acceleration, and multi‑node load balancing, and provides estimates of server counts and bandwidth capacities.
The article explains ByteDance's enormous data‑center bandwidth, server counts, and CDN architecture that enable hundreds of millions of concurrent users on Douyin and TikTok, detailing estimates of total outbound capacity, multi‑link designs, and the role of cloud and IDC resources.
Alibaba Cloud’s infrastructure network team’s split‑type OSFP Cage proposal was unanimously approved by the OSFP MSA committee, becoming the first standard supporting liquid‑cooled OSFP cold plates, offering low‑cost, easy‑assembly solutions that address the growing power‑consumption challenges of high‑density AI switches.
The UALink 1.0 specification, driven by AMD, Intel, Broadcom and other industry leaders, introduces an open, low‑latency, high‑bandwidth interconnect that can link up to 1,024 AI accelerators, offering a cost‑effective alternative to Nvidia’s NVLink and reshaping the AI‑HPC market.
At the OFC 2025 50th anniversary in San Francisco, Alibaba Cloud presented cutting‑edge optical interconnect research and solutions for AI computing and modern data‑center networks, highlighted by invited talks, breakthrough demos, and two data‑driven QoT estimation papers co‑authored with Hong Kong Polytechnic University.
This article details the design, implementation, and performance evaluation of a user‑space MPTCP stack built on DPDK, showing how a layered, zero‑copy architecture and same‑core lock‑free forwarding can boost data‑center throughput by up to 100% while reducing latency by about 10%, all while remaining compatible with existing TCP applications.
NVIDIA’s GTC 2025 keynote outlines the four AI waves—from perception to physical AI—while highlighting the company’s latest Blackwell chips, DGX Spark/Station computers, Dynamo inference accelerator, robotics collaborations, GM autonomous‑driving partnership, and AI‑native 6G efforts, underscoring massive data‑center investment and future challenges.
This article explains how a mixed‑integer programming tool developed by ByteDance's SYS‑DCD team integrates cost, reliability, delivery speed, and environmental metrics to optimize data‑center planning, reduce power waste, and accelerate deployment across multiple regional scenarios.
This article explains RDMA technology, covering its core principles, programming model with Verbs API, various communication modes, and its impact on data‑center networking, high‑performance computing, and distributed storage, highlighting its low‑latency, zero‑copy advantages over traditional TCP/IP.
This article provides a detailed overview of Ethernet’s origins, its rapid speed evolution, the expanding role of Ethernet in data centers, automotive, industrial and cloud networks, and examines the current switch market, chip technologies, and future high‑speed developments driven by AI and AIGC.
The 2024 OCP Global Summit in San Jose showcased Chinese tech leaders like Alibaba Cloud and ByteDance presenting cutting‑edge AI network architectures, liquid‑cooling solutions, SRv6 deployments, high‑performance data‑center designs, and future WAN routing innovations, underscoring China's growing influence in AI infrastructure worldwide.
The article surveys the network architectures and congestion‑control techniques used in massive GPU clusters—such as Byte’s megascale, Baidu HPN, Alibaba HPN7, and Tencent Xingmai 2.0—highlighting how high‑bandwidth, low‑latency designs and advanced RDMA technologies enable training of trillion‑parameter multimodal AI models.
This article presents a comprehensive reference design and performance evaluation of a 2U four‑node high‑density server employing full liquid‑cooled cold plates for CPUs, memory, storage, NICs, and power supplies, detailing system architecture, flow design, CFD validation, and future optimization directions.
The article provides an in‑depth technical overview of Ampere’s custom‑core AmpereOne A192‑32X 192‑core ARM server processor, covering its architecture, cloud‑native features, performance comparisons with AMD EPYC and Intel Xeon, cooling design, LGA5964 socket details, and benchmark results from real‑world stress testing.
This white‑paper presents a 2024 reference design and performance evaluation of full‑liquid‑cooling solutions for CPUs, memory, SSDs, PCIe/OCP cards, power supplies and IO boards, detailing architecture, advantages, implementation methods and deployment scenarios for data‑center and telecom applications.
ByteDance won the 2024 Open Compute Best Practice Award for its groundbreaking work in cloud firmware, OpenBMC, VDUSE virtualization, and kernel memory optimizations, illustrating how open‑source collaboration drives more efficient, scalable data‑center infrastructure in the AI era.
This article provides a detailed overview of modern data center architecture, covering physical and IT infrastructure, network topologies such as three‑tier and spine‑leaf, storage solutions like DAS, NAS and SAN, server designs, cloud data‑center components, physical site considerations, and various data‑center deployment models.
This extensive training program covers everything a data center operations engineer needs—from foundational infrastructure management and server hardware maintenance to advanced network configuration, security hardening, monitoring, fault handling, and practical hands‑on skills for real‑world challenges.