zookeeper原理分析大纲

一、设计目标

• 简单(API:create/delete/exists/get data/set data/get children/sync)
• 可复制(内存)
• 有序(tcp/zxid)
• 快速(nio/读写分离)

二、数据结构

三、Znode

• Persistent vs ephemeral
• Sequence
• Watcher
• Stat

e)ACL
四、ACL Permissions

• CREATE
• READ
• WRITE
• DELETE
• ADMIN
• Builtin ACL Schemes

i.world has a single id, anyone, that represents anyone.
ii.auth doesn’t use any id, represents any authenticated user.
iii.digest uses a username:password string to generate MD5 hash which is then used as an ACL ID identity

五、认证auth

• SASL(Kerberos)
• IP
• Digest
• Other plugins

六、存储

• DataTree
• Snapshot
• txnLog

七、运行模式

• Standalone
• Cluster

1.Standalone
2.Cluster
a)Quorum(participate/leader)/Observer/Readonly
b)Status

c)Leader选举

• Major/Hierarchical
• Voting(两两互联，各自统计)

d)Leader确认

• newEpoch/EpochAck
• Sync:DIFF/TRUNC/SNAP
• NEWLEADER/ACK/UPTODATE

e)Leader工作

• PROPOSAL/COMMIT/INFORM
• ACK/PING/REVALIDATE/REQUEST

八、Client
a)命令行

b)zkClient

• 定义了各种listener接口
• serialize接口
• 提供了一些工具及接口

c)Curator

• 动态加载可使用zookeeper服务器
• 超时及重试机制
• Builder封装
• Namespace
• Compress
• background operation