and useful insights to troubleshoot system performance problems. Projects such as BCC, bpftrace, Hubble, Parca, or Pyroscope use eBPF for tracing and monitoring purposes. The team flew and traveled, becoming

started and experiment with Cilium. Getting Started Introduction to Cilium & Hubble What is Cilium? What is Hubble? Why Cilium & Hubble? Functionality Overview Getting Started Guides Installation Observability Kubernetes constructs in policy Endpoint Lifecycle Troubleshooting Monitoring & Metrics Cilium Metrics Hubble Metrics Example Prometheus & Grafana Deployment Metrics Reference Performance & Scalability Tuning Benchmark Scalability Troubleshooting Component & Cluster Health Observing Flows with Hubble Observing flows with Hubble Relay Connectivity Problems Policy Troubleshooting etcd (kvstore) Cluster Mesh Troubleshooting

started and experiment with Cilium. Getting Started Introduction to Cilium & Hubble What is Cilium? What is Hubble? Why Cilium & Hubble? Functionality Overview Getting Started Guides Installation Network Policy Kubernetes constructs in policy Endpoint Lifecycle Troubleshooting Monitoring & Metrics Cilium Metrics Hubble Metrics Example Prometheus & Grafana Deployment Metrics Reference Scalability Limiting Identity-Relevant Evaluation Results Tuning Troubleshooting Component & Cluster Health Observing Flows with Hubble Observing flows with Hubble Relay Connectivity Problems Policy Troubleshooting etcd (kvstore) Symptom Library Useful

Cilium code or documentation. Getting Started Introduction to Cilium & Hubble What is Cilium? What is Hubble? Why Cilium & Hubble? Functionality Overview Getting Started Guides Installation Network Policy Kubernetes constructs in policy Endpoint Lifecycle Troubleshooting Monitoring & Metrics Cilium Metrics Hubble Metrics Example Prometheus & Grafana Deployment Metrics Reference Scalability Limiting Identity-Relevant Evaluation Results Tuning Troubleshooting Component & Cluster Health Observing Flows with Hubble Observing flows with Hubble Relay Connectivity Problems Policy Troubleshooting etcd (kvstore) Symptom Library Useful

experiment with Cilium. Getting Started • Introduction to Cilium & Hubble What is Cilium? What is Hubble? Why Cilium & Hubble? • Functionality Overview • Getting Started Guides ○ Installation Endpoint Lifecycle o Troubleshooting • Monitoring & Metrics • Cilium Metrics • Hubble Metrics • Example Prometheus & Grafana Deployment • Metrics Reference • Performance Troubleshooting • Component & Cluster Health • Observing Flows with Hubble • Observing flows with Hubble Relay • Connectivity Problems • Policy Troubleshooting • etcd

metrics are exported via Prometheus for integration with your existing dashboards. Hubble [https://github.com/cilium/hubble/]: An observability platform specifically written for Cilium. It provides service Subsequent commands reference this environment variable. export CILIUM_NAMESPACE=kube-system Install Hubble Hubble is a fully distributed networking and security observability platform for cloud native workloads manner. Visit Hubble Github page [https://github.com/cilium/hubble]. Generate the deployment files using Helm and deploy it: git clone https://github.com/cilium/hubble.git --branch v0.5 cd hubble/install/kubernetes

73a43f4e091ce5431ae7074958/p9_3.jpg) Hubble UI Flow Inspection  ./jq Hubble CLI  pod Cilium pod BPF ## Evaluating Cilium and Hubble ## Do it, Do it, Do it! ![Image](/uploads/documents/0/2/5/e/025ee573a43f4e091ce5431ae7074958/p10_1 policies ☐ Additional features like IPSec, Cluster Mesh, and more ## Reduced iptables Complexity root@hubble-demo-worker:/# iptables -S -P INPUT ACCEPT -P FORWARD ACCEPT -P OUTPUT ACCEPT —N CILIUM

15 // parameter 16 const type h0 = (argc > 1) ? std::std::vec(argc[1]) : 0.677; 17 const type hubble_0 = h0 * 100 / mpc; 18 const type omega_r0 = (argc > 2) ? std::std::vec(argv[2]) : (2.47E-5 * What is this program doing? 1 // Computation 2 template 3 constexpr auto compute(T hubble_0, T omega_r0, T omega_m0, T omega_10, T omega_k0, T h, T a0) { 4 // Variables 5 T t = 0; evolution.reserve((static_cast(1) / hubble_0) / h); 10 // Function to integrate 11 auto da_dt = [k](auto t, auto a) { 12 return a * hubble_0 * std::sqrt( 13 omega_r0 * std::pow(a

Proxy ## 控制平面 MCP / 私有协议 MCP / 私有协议 MCP / 私有协议 MCP / 私有协议 MOA / Agent 注册中心 服务治理平台 Pangu 配置中心 Hubble 监控平台 • 部署方式 • 升级方式 • 性能问题 • 容灾方式 • 资源问题 • 兼容问题 关键设计 关键问题 ## 容器化运行方式 • sidecar模式 与业务进程相同Pod不同Container ea3517c887fe99a224c71982381f4a7/p23_2.jpg) Service B 业务进程 数据平面 MOA / Agent 注册中心 Pangu 配置中心 Hubble 监控平台 ## 总结与展望 ## 实践感悟 • Service Mesh通过一种全新的架构理念，系统地解决了微服务领域的关键架构痛点 • 是否应用Service Mesh架构以及如何选择

.jpg) 灰度机制 每个App的配置关联一个Task 模板. 配置关联Task模板后，可以服用平台的基于设备、IDC、运营商等多维度的灰度机制 ## 运维平台－系统联动 监控系统 Hubble ↓ Fast运维平台 资产管理平台 天工 智能调度平台 伏羲 ## 运维平台－权限管理 资产平台 Fast Task模板 服务 设备资源 模板(Template) 17c630f8244160/p28_4.jpg) ## 可视化配置(河图) 监控报警配置 根据需求设计，设定业务需求的维度(tag)字段和指标字段(数值类) 投递入河图的数据对接了公司Hubble系统，可以针指标进行配置监控逻辑