node in the figure represents a data node (ChunkServer). A CopySet is the basic unit of data replication in CurveBS. A CopySet contains multiple chunks. Copysets in CurveBS are created in advance. 1. Reduce metadata. If the replication group membership is saved for each chunk, the number of metadata will be large. 2. Example Reduce the number of replication groups. The size of chunk is such as 16MB by default. If replication groups are created based on chunk. The number of replication groups is large and the communication traffic between replication groups is heavy. Once a CopySet

Benchmark TPC-H TiDB DM: TiDB Data Migration Tool TiDB DM (Data Migration) is an integrated data replication task management platform that supports: ● full data migration ● incremental data migration Amazon Aurora ○ https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/mysql-replication-gtid.html#mysql-replication-gtid.co nfiguring-aurora ● Make sure DM can connect to Amazon Aurora master (check security group rules) ● Ignore some unsupported syntax ○ ignore-checking-items: ["dump_privilege", "replication_privilege"] ○ example task.yaml: https://gist.github.com/c4pt0r/bcbf645ab475e5a603ad3d79d95c1757

workload Low-cost Data Replication Region A Region A Region A TiKV TiKV TiKV TiFlash R e g i o n A Leader Follower Follower Learner Low-cost Data Replication ● Although data replication is asynchronous

tests ran on a five-node De- bian Stretch cluster, with PD, TiKV, and TiDB on each node. We used a replication factor of 3 for each TiKV region. To stress cross-region transactions, we used TiDB’s split-table where the num- ber of replicas falls below the configured target. In ad- dition, waiting for full replication before starting TiDB (or any other service which uses TiKV), can dramati- cally speed up convergence the upcoming 3.0.0-rc.3 release; deployment systems could use this flag to ensure full region replication as a part of cluster setup. 3.4 Read Skew & Lost Updates Under normal operation, without faults

[server] 2. labels = "zone=,rack=,host=" 可以通过 PD 的配置文件让 PD 理解 TiKV 集群的拓扑结构。 1. [replication] 2. max-replicas = 3 3. location-labels = ["zone", "rack", "host"] 其中 location-labels 需要与 TiKV binlog 且格式必须为 ROW。 注意： 只有上文提到的第二种场景才需要在 dump 数据之前先开启 binlog MySQL 开启 binlog 功能，参考 Setting the Replication Master Configuration Binlog 格式必须使用 ROW format，这也是 MySQL 5.7 之后推荐的 binlog 格式，可以使用如下语 句打开: 1. 时，可以添加 --no-locks 参数，避免申请 reload 权限 6. 检查上下游同步用户权限 需要上游 MySQL 同步账号至少赋予以下权限： select , replication slave , replication client 下游 TiDB 可暂时采用 root 同权限账号 7. 检查 GTID 与 POS 相关信息 使用以下语句查看 binlog 内容： show binlog

mirroring configured with CRDs • Different Curve clusters may share MDS and ETCD server • Replication between CurveFS configured with CRDsFeature list for CurveBS • CSI plugin for CurveBS • CurveBS

• GTID • 多流业务系统 上游 Single Source of Truth 问题背景 已有系统迁移怎么办？ 下游Demo for ReplicationPolarDB-X Replication 特性详情 提供与 MySQL 主备复制的能力 产品体验 • 支持 MySQL Change Master 指令 • 原生作为 MySQL 备库的能力 • 支持 PolarDB-X