https://www.chenshaowen.com/en/tags/distributed-systems/Recent content in Distributed Systems on Shaowen Chen's WebsiteHugo -- gohugo.ioen©2016 - {year}, All Rights Reserved.Sun, 08 Feb 2026 00:00:00 +0000weeklyhttps://www.chenshaowen.com/en/blog/multi-node-multi-disk-minio-cluster-performance-test-in-different-erasure-code-configurations-on-ipoib.htmlSun, 08 Feb 2026 00:00:00 +0000Sun, 08 Feb 2026 00:00:00 +0000https://www.chenshaowen.com/en/blog/multi-node-multi-disk-minio-cluster-performance-test-in-different-erasure-code-configurations-on-ipoib.htmlThe earlier tests found the bottleneck in the NIC; this post adds supplementary testing under IPoIB. 1. MinIO Cluster Environment 1.1 Creating the MinIO Cluster 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 export CONTAINER_CLI=nerdctl export IMAGE=minio/minio:RELEASE.微信公众号MinIOStorageDistributed SystemsErasure CodingIPoIBPerformance TestingBenchmarkinghttps://www.chenshaowen.com/en/blog/minio-multi-node-multi-disk-deployment-and-maintenance.htmlSat, 07 Feb 2026 00:00:00 +0000Sat, 07 Feb 2026 00:00:00 +0000https://www.chenshaowen.com/en/blog/minio-multi-node-multi-disk-deployment-and-maintenance.html1. Environment Preparation 1.1 Preparing Data Disks Inspect the data disks 1 lsblk -d -o NAME,SIZE,TYPE | grep nvme 1 2 3 4 nvme0n1 745.2G disk nvme1n1 745.2G disk nvme2n1 745.2G disk nvme3n1 745.2G disk Prepare the storage directories 1 2 3 for i in {0..3}; do mkdir -p /mnt/data${i} done Format the data disks 1 2 3 for i in {0.微信公众号MinIOStorageDistributed SystemsOperationsLearningObject StorageS3DeploymentHigh Availabilityhttps://www.chenshaowen.com/en/blog/what-is-ray.htmlSun, 09 Feb 2025 10:00:00 +0000Sun, 09 Feb 2025 10:00:00 +0000https://www.chenshaowen.com/en/blog/what-is-ray.html1. What Ray Is In 2016, UC Berkeley’s RISELab released a new distributed computing framework called Ray. In 2017, after the Ray paper was published, it drew broad attention across the industry; in China it was mainly Ant Group that adopted and contributed to Ray. In 2020, Ray released version微信公众号Distributed SystemsComputing FrameworkRayTrainingInferencehttps://www.chenshaowen.com/en/blog/mpi-communication-primitives-and-python-programming.htmlSun, 10 Nov 2024 00:00:00 +0000Sun, 10 Nov 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/mpi-communication-primitives-and-python-programming.html1. What MPI Is MPI, the Message Passing Interface, is a communication protocol used for parallel computing. MPI provides a set of standardized interfaces for transferring data between different compute nodes, and is widely used in scientific computing, machine learning, deep learning, and other fields. MPI has multiple implementations; the微信公众号MPIAIDistributed SystemsCommunicationhttps://www.chenshaowen.com/en/blog/data-parallel-architecture-in-distributed-train.htmlWed, 21 Aug 2024 00:00:00 +0000Wed, 21 Aug 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/data-parallel-architecture-in-distributed-train.html1. Parameter Server Architecture In the Parameter Server architecture, the nodes in the cluster are divided into two categories: parameter server nodes (Parameter Server) and worker server nodes (Worker). 1.1 Parameter Server The Parameter Server is used to store the model’s parameters. Each parameter server node is responsible for managing and updating one part of the model’s parameters, while each worker node only processes the subset of data corresponding to it.微信公众号Distributed SystemsTrainingData ParallelismAIOperationshttps://www.chenshaowen.com/en/blog/using-kubefed-to-distribute-tekton-resource-cross-cluster.htmlSat, 26 Jun 2021 00:00:00 +0000Sat, 26 Jun 2021 00:00:00 +0000https://www.chenshaowen.com/en/blog/using-kubefed-to-distribute-tekton-resource-cross-cluster.html1. The Advantages of Multi-Cluster Tekton Builds Thanks to Kubernetes, Tekton already has good elasticity and can support large-scale builds. At the same time, developing Tasks mainly uses Yaml and Shell, which widens the range of scenarios Tekton can adapt to. The figure above is a sketch of Tekton across multiple clusters.微信公众号TektonKubeFedKubernetesMulti-ClusterCICDOperationsLearningPipelineBuildDistributed Systems