Shell 提供强大的界面，有助于自动化和集成所有组件 InnoDB 克隆以自动生成成员，完全集成在InnoDB 中 MySQL Router MySQL Server RPO =0 RTO =seconds ( 自动故障转移) Copyright @ 2021 Oracle and/or its affiliates. 9 / 55 Present - Solutions Replicaset • "经典"、"异步"的基于复制的解决方案，完全集成 • MySQL Shell • MySQL Router • MySQL Server RPO !=0 RTO =minutes (manual failover) Copyright @ 2021 Oracle and/or its affiliates. MySQL InnoDB ClusterSet 一个或者多个MySQL InnoDB Cluster 副本连接到一个主MySQL InnoDB Cluster 高可用 (区域内的失败) RPO =0 RTO =秒级 (自动故障转移) 灾难恢复（ 区域故障） RPO !=0 RTO =分钟或更长时间（ 手动故障转移） 无写入性能影响 特点 • 简单易用 • 熟悉的界面和可用性 mysqlsh,CLONE,... • 在线添加/删除节点/集群

⽅案存 在以下缺点：资源利⽤率低、维护成本⾼、RTO (Recovery Time Objective) 及 RPO (Recovery Point Objective) ⽆法真实达到企业所期望的值。TiDB 采⽤多副本 + Multi-Raft 协议 的⽅式将数据调度到不同的机房、机架、机器，当部分机器出现故障时系统可⾃动进⾏切换，确保系统的 RTO <= 30s 及 RPO = 0。 对存储容

NewSQL | TiDB 简单配置即可保障所有的数据中心 100% 同步，真正意义上的多活 完全实现 国标（信息安全技术-信息系统灾难恢复规范GBT 20988-2007 ）最高等级 第六级的要求（RTO = 0，RPO = 0） NewSQL - 大数据量下高压力实时处理的完美解决方案 随着业务增长，需要处理的数据量不断增加，系统架构师面临多种挑战： ● 传统的关系型数据库无法满足高并发写入？

solution has the disadvantages of low resource utilization, high maintenance cost, and the fact that RTO (Recovery Time Objective) and RPO (Recovery Point Objective) cannot meet expectations. TiDB uses multiple and machines. When some machines fail, the system can automatically switch to ensure that the system RTO ￿ 30 seconds and RPO = 0. • Massive data and high concurrency scenarios with high requirements for one TiKV can reach 9 MiB/s, which is 50% faster than before. The restore speed is scalable and the RTO in DR scenarios is reduced greatly. The RPO in DR scenarios can be as short as 5 minutes. In normal

Multi-Raft protocol to schedule data to different data centers, racks, and machines, ensuring system RTO ￿ 30 seconds and RPO = 0. • Massive data and high concurrency scenarios Traditional standalone databases Describes the DR solutions provided by TiDB. • Compares these DR solutions. 8.4.1.1 Basic concepts • RTO (Recovery Time Objective): The time required for the system to recover from a disaster. • RPO (Recovery business can tolerate in a disaster. The following figure illustrates these two concepts: Figure 112: RTO and RPO • Error tolerance objective: Because a disaster can affect different regions. In this doc-

Multi-Raft protocol to schedule data to different data centers, racks, and machines, ensuring system RTO ￿ 30 seconds and RPO = 0. • Massive data and high concurrency scenarios Traditional standalone databases Describes the DR solutions provided by TiDB. • Compares these DR solutions. 8.4.1.1 Basic concepts • RTO (Recovery Time Objective): The time required for the system to recover from a disaster. • RPO (Recovery business can tolerate in a disaster. The following figure illustrates these two concepts: Figure 112: RTO and RPO • Error tolerance objective: Because a disaster can affect different regions. In this doc-

Multi-Raft protocol to schedule data to different data centers, racks, and machines, ensuring system RTO ￿ 30 seconds and RPO = 0. • Massive data and high concurrency scenarios Traditional standalone databases Describes the DR solutions provided by TiDB. • Compares these DR solutions. 8.4.1.1 Basic concepts • RTO (Recovery Time Objective): The time required for the system to recover from a disaster. • RPO (Recovery business can tolerate in a disaster. The following figure illustrates these two concepts: Figure 112: RTO and RPO 932 • Error tolerance objective: Because a disaster can affect different regions. In this

在以下缺点： 资源利用率低、维护成本高、RTO (Recovery Time Objective) 及 RPO (Recovery Point Objective) 无法真实达到企业 30 所期望的值。TiDB 采用多副本 + Multi-Raft 协议的方式将数据调度到不同的机房、机架、机器，当部分机 器出现故障时系统可自动进行切换，确保系统的 RTO <= 30s 及 RPO = 0。 • [@joccau](https://github.com/joccau) PITR 恢复的日志恢复阶段，单台 TiKV 的恢复速度可以达到 9 MiB/s，提升了 50%，并且恢复速度可扩展，有效 地降低容灾场景的 RTO 指标；容灾场景的 RPO 优化到 5 分钟，在常规的集群运维，如滚动升级，单 TiKV 故障 等场景下，可以达到 RPO = 5 min 的目标。 38 • TiKV-BR 工具 GA，支持 RawKV 对比不同的容灾解决方案。 686 8.4.1.1 基本概念 • RTO (Recovery Time Objective)：是指灾难发生后，系统恢复服务所需的时间。 • RPO (Recovery Point Objective)：是指灾难发生后，确保对业务不产生损失的前提下，可以丢失的最大数据 量。 下面的图形描述了这两个概念： 图 54: RTO and RPO • 错误容忍目标：由于灾难可能影响的地域范围是不同的，在本文中，使用

金融行业对数据一致性及高可靠、系统高可用、可扩展性、容灾要求较高。传统的解决方案的资源利用 率低，维护成本高。TiDB 采用多副本 + Multi-Raft 协议的方式将数据调度到不同的机房、机架、机器，确 保系统的 RTO <= 30s 及 RPO = 0。 31 • 海量数据及高并发的 OLTP 场景 传统的单机数据库无法满足因数据爆炸性的增长对数据库的容量要求。TiDB 是一种性价比高的解决方 案，采用计算 对比不同的容灾解决方案。 8.4.1.1 基本概念 • RTO (Recovery Time Objective)：是指灾难发生后，系统恢复服务所需的时间。 • RPO (Recovery Point Objective)：是指灾难发生后，确保对业务不产生损失的前提下，可以丢失的最大数据 量。 733 下面的图形描述了这两个概念： 图 57: RTO and RPO • 错误容忍目标：由于灾难可 两个集群之间通过 TiCDC 进行数据改变的同步。这种架构，简称为 “1:1” 解决方案。 这种架构看起来非常简洁，可用性比较高，最大的错误容忍目标可以做到区域级别，写能力也能够得到扩展， RPO 在秒级别，RTO 在分钟级别，甚至更低。如果 RPO 为 0 并不是必须满足的要求，推荐在重要生产系统使用 该容灾方案。对于该方案的详细信息，请参考基于主备集群的容灾方案。 8.4.1.3.2 基于多副本的单集群容灾方案

Multi-Raft protocol to schedule data to different data centers, racks, and machines, ensuring system RTO ￿ 30 seconds and RPO = 0. • Massive data and high concurrency scenarios Traditional standalone databases Describes the DR solutions provided by TiDB. • Compares these DR solutions. 8.4.1.1 Basic concepts • RTO (Recovery Time Objective): The time required for the system to recover from a disaster. • RPO (Recovery business can tolerate in a disaster. The following figure illustrates these two concepts: Figure 112: RTO and RPO 940 • Error tolerance objective: Because a disaster can affect different regions. In this