zookeeper系列(七) 官方翻译简介

Performance

ZooKeeper is designed to be highly performant. 

But is it? The results of the ZooKeeper's development team at Yahoo! Research indicate that it is. 

(See ZooKeeper Throughput as the Read-Write Ratio Varies.) 

It is especially high performance in applications where reads outnumber writes, 

since writes involve synchronizing the state of all servers. 

(Reads outnumbering writes is typically the case for a coordination service.)

The figure ZooKeeper Throughput as the Read-Write Ratio 

Varies is a throughput graph of ZooKeeper release 3.2 running on 

servers with dual 2Ghz Xeon and two SATA 15K RPM drives. 

One drive was used as a dedicated ZooKeeper log device. 

The snapshots were written to the OS drive. Write requests were 1K writes and the reads were 1K reads. 

"Servers" indicate the size of the ZooKeeper ensemble, 

the number of servers that make up the service. 

Approximately 30 other servers were used to simulate the clients. 

The ZooKeeper ensemble was configured such that leaders do not allow connections from clients.

性能测试: zookeeper 被设计的号称高性能框架,但是事实情况如何呢?

  来自雅虎的zookeeper开发团队给我们带来了结果.研究结果表明事实确实如此.

  (可以查看下zookeeper 吞吐量随着读写比例变化的图表)

  在读占绝对优势的应用中,性能尤佳,因为写操作涉及到服务器之间状态的同步.尤其是在协调服务这个典型案例中表现的尤其突出.

   

  zookeeper 吞吐量和读写比例的变化关系用的是zookeeper3.2版本,运行在 双核 2Ghz 及SATA 15k RPM 处理器配置的服务器中.

  其中一个负责zookeeper 日志设备. 快照信息写入操作系统驱动中.

  写入请求是1Kb写入, 读请求也是1kb的写入(读写单元都是1Kb).

  servers 标示着 整个zookeeper的集群大小, 即组成zookeeper服务的服务器数.

  整个zookeeper集群有个限定,客户端都不能直接连接zookeeper的leader 节点.