GPU Resource Management On JDOS 梁永清 liangyongqing1@jd.com 提供的服务 1. 用于实验的 GPU 容器 2.基于 Kubeflow 的机器学习训练服务 3.模型管理和模型 Serving 服务 Experiment Training Serving 均基于容器，不对业务方直接提供 GPU 物理机 GPU 实验 JDOS 常规的容器服务

Secrets Management at Scale with Vault & Rancher 24. June Robert de Bock Senior DevOps Engineer Adfinis robert.debock@adfinis.com Kapil Arora Senior Solution Engineer HashiCorp kapil@hashicorp.com Infrastructure Management (Run & Manage) GitOps Continuous Delivery Cluster Templates & Config Enforcement K8s Version Management Node Pool Management Cluster Provisioning & Lifecycle Management Platform Amazon EKS Azure AKS Google GKE Cloud Datacenter Edge Branch Dev Secret Management in Kubernetes 16 17 18 Secret Management Challenges ● Secrets sprawl ● Secrets rotation ● X.509 certificates, SSH

Node Operator: Kubernetes Node Management Made Simple 陈俊(Joe), Ant Financial Agenda • Background and Motivation • Introduction of Operators • Node-Operator • Advanced Topic: • Upgrade Master & Node Components reliably • Canary Rollout • Master & Node Component Versions Management Motivation: Work Order Deployment Worker Order • Upgrade Nodes Versions • Upgrade Node 10.10 • Non-failure-aware • Complicated architecture Work order deployment system can not meet the requirements of resource management. Operator Observe Action Analyze • Observe: watch desired resource

Service Mesh的延伸 之论道Database Mesh 分享人：张亮 日期：2018年07月25日Service Mesh风头正劲Service Mesh产品多样化Service Mesh的优势 云原生 零入侵 可观察性 面向运维服务化之后，数据库怎么办？ 服务 • 无状态 • 根据规则路由 • 业务方处理事务 数据库 • 有状态 • 根据SQL路由 • 数据库自动处理事务数据库的进化趋势 • SQL • ACID+BASE • 分布式 NewSQLNewSQL的分类 New Architecture Transparent Sharding Middleware Database-as-a-Service What's Really New with NewSQL?数据库中间层的优势 系统 •事务 运维 • DBA 开发 • SQL数据库中间层应具备的能力 Sidecar 数据库 任意 单一 单一 连接数 高 低 高 异构语言 仅Java 任意 任意 性能 损耗低 损耗略高 损耗低 无中心化 是 否 是 静态入口 无 有 无 Sidecar的优势Database Mesh架构图Sharding-Sphere 核心功能 数据分片 分布式事务 数据库治理 弹性伸缩 管控界面 实现方案 Sharding-JDBC Sharding-Proxy Sharding-Sidecar

Processing and Analytics Vasiliki (Vasia) Kalavri
 vkalavri@bu.edu Spring 2020 2/25: State Management Vasiliki Kalavri | Boston University 2020 Logic State <#Brexit, 520> <#WorldCup, 480> types • The system is unaware of which parts of an operator constitute state Streaming state 3 • Explicit state primitives including state types and interfaces • The system is aware of state key of the current record so that all records with the same key access the same state State management in Apache Flink 5 Vasiliki Kalavri | Boston University 2020 Operator state Keyed state State

4.3.2. Stop 4.3.3. Status 4.3.4. Restart 4.3.5. SystemMBean 4.4. Integration in the operating system 4.4.1. Service Wrapper 4.4.2. Service Script Templates 4.5. Using the console 4.5.1. Available JMX LogMBean 4.7.6. Advanced configuration 4.8. Configuration 4.8.1. Environment Variables & System Properties 4.8.2. Files 4.8.3. config:* commands 4.8.4. JMX ConfigMBean 4.9. Artifacts repositories 4.14.7. Security providers 4.15. Docker 4.15.1. Docker images 4.15.2. Docker feature 4.15.3. System-wide information 4.15.4. Show the Docker version information 4.15.5. Search image 4.15.6. Pull

Java Database Connectivity(JDBC) interface over thrift, or an Open Database Connectivity (ODBC) interface over a JDBC-to-ODBC bridge to communicate with Kyuubi. RESTful APIs It provides system management single point of failures. It helps achieve zero downtime for planned system maintenance Failure detectability Failures and system load of kyuubi server and engines are visible via metrics, logs, and Authorization Kinit Auxiliary Service Hadoop Credentials Manager Monitoring 1. Monitoring Kyuubi - Logging System 2. Monitoring Kyuubi - Server Metrics 3. Trouble Shooting Tools Kubernetes Tools Spark Block Cleaner

create the Spark application; 2) a user account can only access data and metadata from a storage system, e.g. Apache Hadoop HDFS [https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign Structured Query Language (SQL) and Java Database Connectivity (JDBC) to handle massive data. It helps you focus on the design and implementation of your business system. Run Anywhere Kyuubi can submit Spark Authorization With strong authentication and fine-grained column/row level authorization, Kyuubi keeps your system and data secure. High Availability Kyuubi provides both high availability and load balancing solutions

create the Spark application; 2) a user account can only access data and metadata from a storage system, e.g. Apache Hadoop HDFS [https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign Structured Query Language (SQL) and Java Database Connectivity (JDBC) to handle massive data. It helps you focus on the design and implementation of your business system. Run Anywhere Kyuubi can submit Spark Authorization With strong authentication and fine-grained column/row level authorization, Kyuubi keeps your system and data secure. High Availability Kyuubi provides both high availability and load balancing solutions

create the Spark application; 2) a user account can only access data and metadata from a storage system, e.g. Apache Hadoop HDFS [https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign Structured Query Language (SQL) and Java Database Connectivity (JDBC) to handle massive data. It helps you focus on the design and implementation of your business system. Run Anywhere Kyuubi can submit Spark Authorization With strong authentication and fine-grained column/row level authorization, Kyuubi keeps your system and data secure. High Availability Kyuubi provides both high availability and load balancing solutions