预览： Preview: Oracle 自治数据库 · 预览：Preview: Oracle 自治数据 ......

Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |

预览： Preview:Oracle 自治数据库Oracle Autonomous Database

胡维 Wei Hu副总裁 Vice President

数据库系统技术 Database Systems Technologies


The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, timing and price of any features or functionality described for Oracle's products may change and remains at the sole discretion of Oracle Corporation. Fees apply for new Database product offerings.

Safe Harbor Statement

2


Oracle自治数据库愿景Oracle’s Vision for Autonomous Database

•自治驾驶 Self-Driving

–用户定义服务级别，数据库确保其实现 User defines service levels, DB makes them happen

•自治安全 Self-Securing

–保护免受恶意内部用户和外部用户攻击 Protection from both external attacks and malicious internal users

•自治修复 Self-Repairing

–自动保护避免所有停机 Automated protection from all downtime

3

自治数据库


自治数据库目标 Autonomous Database Goals

• 更少的人工、成本、错误 Much Less Labor, Costs, Errors

• 更高的安全性、可靠性 Much More Secure, Reliable

• 对于开发人员来说，非常容易使用工业强度数据库开展工作 Much easier for developers to get started with an industrial strength DB

• 支持大型企业关键业务工作负载 Support Mission Critical workloads at the biggest enterprises

• 运行于Oracle公有云或Cloud at Customer Run in Public Cloud or Cloud at Customer

• 四年的努力，目前正Beta测试 Four year effort that is in Beta test now

4

自治数据库


Oracle Database 9i, 10g

• 自动存储管理（ASM） Automatic Storage Management (ASM)

• 自动内存管理 Automatic Memory Management

• 自动数据库诊断监视 Automatic DB Diagnostic Monitor (ADDM)

• 自动工作负载信息资料库（AWR） Automatic Workload Repository (AWR)

• 自动Undo表空间 Automatic Undo tablespaces

• 自动段空间管理 Automatic Segment Space Management

• 自动统计信息收集 Automatic Statistics Gathering

• 自动备用管理（代理） Automatic Standby Management (Broker)

• 自动查询重写 Automatic Query Rewrite

Oracle Database 11g, 12c

• 自动SQL调优 Automatic SQL Tuning

• 自动工作负载回放 Automatic Workload Replay

• SQL监视自动捕获 Automatic Capture of SQL Monitor

• 自动数据优化 Automatic Data Optimization

• 自动存储索引 Automatic Storage Indexes

• 自动列缓存 Automatic Columnar Cache

• 自动诊断框架 Automatic Diagnostic Framework

• 自动数据库克隆更新 Automatic Refresh of Database Cloning

• 自动健康框架 Autonomous Health Framework

5

自治数据库之旅 Journey to Autonomous Database

• 几十年来，甲骨文一直在开发复杂的数据库自动化系统 Oracle has been developing sophisticated database automation for decades



• 通过工程化系统，Oracle自动化数据库基础设施 With Engineered Systems, Oracle automated DB infrastructure

6



• 云使Oracle能够交付完全自治的数据库 Cloud enables Oracle to deliver a Fully Autonomous DB

–扩展的数据库自动化 Expanded DB Automation

–集成了完整的基础设施自动化 Integrated with complete infrastructure automation

–还有额外的运营自动化，HA，安全等 With additional automation for operations, HA, security, etc.

7

世界上领先的数据库现在是世界上巨简单的数据库World’s Best Database is Now World’s Simplest

自治数据库


传统上，DBA负责: Traditionally DBAs are Responsible for:

8

• 通用任务 Generic Tasks

– 系统、网络、存储的配置和调优 Configuration and tuning of systems, network, storage

– 数据库供应，打补丁 DB provisioning, patching

– 数据库备份，H/A，灾难恢复 DB backups, H/A, disaster recovery

– 数据库优化 DB optimization

• 特定业务任务 Tasks Specific to Business

– 架构，规划和数据建模 Architecture, planning, data modeling

– 数据安全和数据生命周期管理 Data security and data lifecycle management

– 应用程序关联调优 Application related tuning

– 端到端服务级别管理 End-to-End service level management


• 通用任务 Generic Tasks

– 系统、网络、存储的配置和调优 Configuration and tuning of systems, network, storage

– 数据库供应，打补丁 Database provisioning, patching

– 数据库备份，H/A，灾难恢复 Database backups, H/A, disaster recovery

– 数据库优化 Database optimization

• 特定业务任务 Tasks Specific to Business

– 架构，规划和数据建模 Architecture, planning, data modeling

– 数据安全和数据生命周期管理 Data security and data lifecycle management

– 应用程序关联调优 Application related tuning

– 端到端服务级别管理 End-to-End service level management

自治数据库消除了通用任务 Autonomous Database Removes Generic Tasks

9

DBA从繁杂事务中解放出来，拥有了更多时间去创新、提升业务

自治数据库


议程 Agenda

完全的端到端自动化 Full End-to-End Automation

特定的工作负载 Specialized by Workload

数据库管理员的角色 Role of Database Administrator

1

2

3

10



• 自治数据库必须更安全、更高可用 Autonomous DB must be MORE Secure and MORE Available

– 数据库是至关重要的 – 没有人想让自己自动破产 DBs are critical - no one wants to automate themselves into bankruptcy

• 是什么让这一切成为可能： What makes this possible:

– 专业化 – Oracle 有成千上万的工程师，可以熟练地自动化所有领域 Specialization - Oracle has thousands of engineers, can expertly automate all areas

– 标准化 – 云规模化使规模化学习，规模经济成为可能 Standardization - Cloud scale enables learning at scale and economies of scale

• 自动化是旅程 – 像其它所有技术一样 Automation is a Journey – like every other technology

– 初始版本自动化非常完整 Initial release automation is very complete

– 自动化将不断变得越来越聪明，越来越好 Will keep getting smarter and better

11

自治数据库


自动 vs. 自治 Automated vs. Autonomous

• 客户可以选择仅使用自动化或将所有管理移交给Oracle云运营进行自治运营 Customer can choose to just use automation or hand over all management to Oracle Cloud Operations for Autonomous operation

• 如果客户将管理移交给Oracle，那么: If Customer hands over management to Oracle then:

–无需提供数据库和OS管理员权限 DB and OS Administrator Privileges are not needed and not provided

–异常和失败案例由Oracle专家处理 Exception and failures are handled by Oracle experts

• 回报是巨大的 – 消除通用任务，减少人工，降低成本，减少错误，同时增加安全性和可用性 Payoff is huge - eliminate generic tasks, reduce labor, reduce costs, reduce errors, while increasing security and availability

12


自治数据库自动供应Autonomous Database is Automatically Provisioned

• 预构建，预优化，预测试 Pre-built, pre-optimized, pre-tested

13


自动供应 Automatically Provisioned

• 自动供应一个可伸缩、高可用的数据库 Automatically provisions a scalable, highly available DB

– 查找、分配和配置所有需要的硬件和软件 Find, allocate, and configure all needed hardware and software

– 为客户或工作组提供私有的软件定义网络 Private software defined network for customer or workgroup

– 服务器，存储，虚拟机 Server, Storage, Virtual Machine

– Oracle Database 18c 容器或可插拔数据库 Oracle Database 18c Container or Pluggable Database

– 可伸缩的RAC集群 Scalable RAC Cluster

– 远程备用数据库与第三区域的观察者进行自动故障转移 Remote Standby Database with Observer at third region for automatic failover

• 最初的版本运行在Exadata公共云和Exadata Cloud at Customer Initial releases run on Exadata Public Cloud and Exadata Cloud at Customer

14

Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 15

自治数据库自动在线打补丁Autonomous Database Automatically Patches Itself Online


自动打补丁 Automatically Patches Itself

• 一季度一次全堆栈打补丁 Full stack patching once a quarter

– 固件，OS，VM，Clusterware，数据库，使用预构建的Gold映像 Firmware, OS, VM, Clusterware, Database, using prebuilt Gold Image

– 打补丁前运行预先检查，以确保可靠性 Pre-checks run in days before patching to ensure reliability

• 跨集群节点的滚动式应用，以获得可用性 Applied in a rolling fashion across nodes of cluster for availability

• 客户指定维护窗口 Customer specifies maintenance window

– 打补丁的时间自动选定，客户可以覆盖 Time to apply patch is automatically selected, customer can override

• Oracle云每天为成千上万的系统打补丁 Oracle Cloud patches thousands of systems in a day

– 打补丁的规模极大地提高了自动化和可靠性 Scale of patching has driven huge improvements in automation and reliability

16


自治数据库自动保护自己Autonomous Database Automatically Secures Itself


自动保护自己 Automatically Secures Itself

• 自动保护免受外部攻击 Automated Protection from external attacks

– 自动在线应用最新的安全更新 Automatically applies latest security updates online

– 原生加密可以防止数据库外部的数据访问 Native encryption prevents data access from outside the database

• 允许由客户现有的安全模块(HSM)管理密钥 Allows keys to be managed by customer’s existing Security Module (HSM)

• 将结合Database Aware Key Management云服务 Will interface with DB Aware Key Management Cloud Service

• 自动保护抵御恶意的内部用户 Automated Protection from malicious internal users

– Database Vault防止管理员看到用户数据 Database Vault prevents administrators from seeing user data

– Data Masking发现和遮蔽测试人员的敏感数据 Data Masking finds and conceals sensitive data from testers

• Oracle员工被锁定在“破碎玻璃”程序中，有客户的批准，Oracle运维才能够访问系统 Oracle staff is locked out - well defined “break glass” procedure, with customer approval, for Oracle Operations to access system

18


自治数据库自动监控自己Autonomous Database Automatically Monitors Itself


自动监控自己 Automatically Monitors Itself

• 监控关键部件的容量限制和瓶颈 Monitors for capacity limits and bottlenecks in key components

– CPU，内存，存储，网络 CPU, Memory, Storage, Network

– 确保资源枯竭不会导致性能低下或中断 Ensures resource exhaustion does not cause poor performance or outage

– 警报阈值因应用程序而异，例如OLTP vs. 数据仓库 Alert thresholds vary by application, e.g. OLTP vs Warehouse

• 资源管理阻止一个用户或数据库独占资源 Resource Management prevents one user or database from monopolizing resources

• 集群健康监视器使用机器学习监视活动，识别异常事件，并推荐(或自动执行)纠正措施 Cluster Health Monitor uses Machine Learning to monitor activity, recognize anomalous events and recommend (or automatically implements) corrective actions

20


自治数据库自动伸缩Autonomous Database Automatically Scales


自动弹性和伸缩 Automated Elasticity and Scaling

• 独立的计算和存储伸缩 Independent scaling of compute and storage

–在线添加和删除cpu或整个计算或存储服务器 Add and remove CPUs or entire compute or storage servers completely online

• 在客户定义范围内即时自动伸缩 Instantly auto-scales within customer defined limits

• 弹性消除了过度供应，降低了成本 Elasticity eliminates overprovisioning, reducing costs

22


自治数据库自动备份Autonomous Database Automatically Backs Itself UP


自动备份 Automatically Backed Up

• 备份计划在夜间进行 Backups are scheduled on nightly basis

• 备份到数据库备份云服务 To Database Backup Cloud Service

• 或未来的零数据丢失恢复云服务，以获取 Or Zero Data Loss Recovery Cloud Service in the future for

• 备份的保留时间是可配置的 Retention Time for backups is configurable

24


自治数据库自动抵抗失败Autonomous Database Automatically Withstands Failures


自动抵抗失败 Automatically Withstands Failures

• 自治数据库必须比手动数据库安全 Autonomous DB must be safer than manually operated DB

• 超常的SLA保证： Exceptional SLA Guarantee:

– 99.995 NRX% (No Ridiculous eXclusions)

– 如果一个月内停机时间多于2.5分钟，保证触发 Guaranteed triggered if there is more than 2.5 minutes downtime a month

– 包括打补丁和升级，数据库bug，区域性中断 Includes patching and upgrade, database bugs, regional outages

• 使用Oracle最佳实践构建了HA的应用程序通常会看到暂停，而不是失败Apps that are architected for HA will usually see pauses, not failures

26

中断 Outage 主要功能KeyFeature

可能的停机时间PotentialDowntime

服务器中断(HA) Server Outage (HA)

RAC 接近于零 Near-Zero

区域性中断，灾难恢复Regional Outage,Disaster Recovery

ADG 数秒 Seconds

数据损坏 Data Corruption ADG 零 Zero

补丁（更新）Patches (Updates) RAC 接近于零 Near-Zero

数据库升级 Database Upgrade ADG 数秒 Seconds

表/索引变更Table/Index Changes Redef 零 Zero

用户错误 User Error Flashback 时间取决于错误 TimeSince Error


自治数据库自动诊断性能Autonomous Database Automatically Diagnoses Performance


自动诊断性能 Automatically Diagnoses Performance

• 自治数据库包括Oracle的行业领先诊断自动化 Autonomous Database includes Oracle’s industry leading diagnostics automation

• 自动数据库诊断监视器（ADDM） Automatic Database Diagnostic Monitor (ADDM)

–自动诊断性能问题的根因 Automatically diagnoses root cause of performance issues

– A.I.（专家系统） A.I. (Expert System)

• 活动工作负载资料档案库（AWR） Active Workload Repository (AWR)

–自动保留详细的性能和资源利用历史 Keeps detailed performance and resource utilization history

• 实时SQL监控 Real-Time SQL Monitoring

–自动诊断如何在SQL语句中使用资源 Automatically diagnoses how resources are used in SQL statements

28


自治数据库自动优化Autonomous Database Automatically Optimizes Itself


自动优化 Automatically Optimizes Itself

• 自治数据库包括Oracle的行业领先数据库优化自动化 Autonomous DB includes Oracle’s industry leading database tuning automation

• 许多数据库算法自我优化 – 缓存、锁定、存储索引、卸载等 Many DB algorithms self optimize – caching, locking, storage indexes, offload..

• 随着新数据的加载，优化器现在通过收集统计数据进一步自动化 Optimizer automatically gathers statistics as new data is loaded

• 自动SQL(Re)调优 Automatic SQL (Re)Tuning

– 机器学习技术，基于新鲜的统计数据不断重新评估SQL计划并建议/实现更好的计划 Machine learning technology that is constantly re-evaluating SQL plans based on the latest statistics and recommending/implementing better plans

• 优化是与工作负载相关的，例如OLTP vs. 分析，因此我们指定服务 Tuning is workload dependent – e.g. OLTP vs analytics so we specialize services

• 优化是一个极其困难的问题 Tuning is an extremely difficult problem

– 即使安排一辆卡车来优化运输也是非常复杂 Even scheduling a fleet of trucks to optimally make deliveries is incredibly complex

– 数据库具有许多必须考虑的自由度和权衡 Database has many degrees of freedom and tradeoffs that must be considered

30


自治数据库自动处理错误Autonomous Database Automatically Handles Errors


自动处理错误 Automatically Handles Errors

• 使用关键的诊断工具扫描堆栈所有层的问题 Scans for issues across all layers of stack using key diagnostic tools

– ORAchk, EXAchk, oswatcher, procwatcher等 ORAchk, EXAchk, oswatcher, procwatcher, etc.

• 如果出现错误： If an error occurs:

– 收集所有相关的诊断数据 Collects all relevant diagnostic data

– 分析日志，建立一个基线，以区分最初的问题和附带后果 Analyses logs to establish a timeline to distinguish initial problem from side-effects

– 上传到Oracle Support Uploads to Oracle Support

– 机器学习算法分析诊断与已知原因相匹配 Machine Learning algorithm analyses diagnostics to match to known cause

• Oracle工程使用机器学习来防止问题发生 Oracle Engineering using Machine Learning to prevent issues from happening

– 预期客户可能会碰上的bug，并优先修复这些缺陷 Anticipate bugs customers are likely to hit, and prioritize fixing those first

– 确定回归测试开发人员在修改代码时应该首先运行 Determine regression tests to run first when modifying code

32


议程 Agenda




1

2

3

33


自治数据库 – 基于用例优化One Autonomous Database – Optimized by Use Case

34

Oracle自治数据库Oracle Autonomous Database

企业级OLTP，混合负载

EnterpriseOLTP,

Mixed Workloads

数据仓库Data Warehousing

部门，开发者Departments,

Developers

2017 2018 现在 Now


议程 Agenda




1

2

3

35


数据库管理员的疑问和恐惧Database Administrator Questions and Fears

•我的工作会消失吗？ Will my job go away?

•我的工作会改变吗? Will my job change?

•我将失去控制权? Will I lose control?

36


现实 vs. 恐惧 Reality vs. Fears

• 您的工作不会消失 – 缺乏熟练的数据库专家 Your job will not go away - there is a shortage of skilled DB experts

– 数据库自动化已经改善了几十年 DB automation has been improving for decades

• 但您的工作将会改变，所以您必须改变 Your job will change, so you must change

• 你会花更少的时间在一般的维护上，花更多的时间去创新 You will spend less time on genericmaintenance, more time innovating

• 业务上花更多的时间 More time with the business

– 做更多的项目，减少积压，从数据中获得更多的价值 Do more projects, get more value from data

• 与开发者花更多时间 More time with developers

– 优化数据访问，提高最终用户体验 Optimizing data access, improving end-user experience

• 花更多的时间在新技术上，如：机器学习 More time on new techniques like Machine Learning

37


数据库管理员必须确保端到端服务级别Database Administrator Must Ensure End-to-End Service Levels

• 专注于优化应用程序和数据库如何协同工作 Focus on optimizing how apps and DBs work together:

– 定义良好的数据模型和良好的SQL Define good data model, and good SQL

– 避免行一次处理，重复登录/解析 Avoid row at a time processing, and repeated logins/parsing

– 了解并行、计划、索引、分区等的权衡 Understand tradeoffs in parallelism, plans, indexes, partitions, etc.

– 确保对敏感数据进行端到端保护 Ensure that sensitive data is kept secure end-to-end

– 请注意应用程序要求什么 – 数据库将完全做到这一点 Beware what an app asks for – DB will do exactly that

• 需要云系统，云网络，云存储方面的新技能 Need new skills in Cloud Systems, Cloud Networking, Cloud Storage

– DBA做容量规划，监视，确保完整的堆栈按预期工作 DBA sizes, monitors, ensures full stack works as expected

• 数据库管理员将获得更多的控制权 Database Administrator will gain more control

– 将负责端到端服务级别的控制 Will be in charge of, and in control of, end-to-end service levels

38


总结：Oracle自治数据库Summary: Oracle Autonomous Database

• 世界上第一个自治关键业务数据库 World’s First Autonomous Mission Critical DB

– 自治驾驶，自治安全，自治修复 Self-Driving, Self-Securing, Self-Repairing

– 完全的端到端自动化 Full End-to-End Automation

– 更少人工，更低成本，更少错误，更高安全性，更可靠 Less Labor, Lower Cost, Fewer Errors, More Secure, More Reliable

– 将DBAs从乏味的通用操作中解放出来，以进行更多的创新 Frees DBAs from generic

operational drudgery to innovate more

• 运行 Runs

– 世界第一数据库，特定的工作负载 World’s #1 Database, Specialized by Workload

– 在公有云或Cloud at Customer In Public Cloud or Cloud at Customer

39

自治数据库

预览： Preview: Oracle 自治数据库 · 预览：Preview: Oracle 自治数据 ......

Documents

Transcript of 预览： Preview: Oracle 自治数据库 · 预览：Preview: Oracle 自治数据 ......