By

about zeromq

大神已经走了，留下的才更显弥足珍贵.

ZeroMq 作者

ZeroMq Manual

ZeroMq guide

ZeroMq API

神器篇之初识 ZeroMQ

why zeromq

raw socket 在面临海量数据和并行时的一些问题，正好是 zeromq 这类 message oriented middleware 要解决的:

  1. block IO or non-block IO? 前者低效，后者复杂;
  2. 传统 socket 的 cs 定义，在实际中 server client 随时可能掉线，特别是 server 挂掉很要命;
  3. socket 传输的数据有大有小，五花八门，raw socket 能 handle?
  4. 有些复杂的应用场景，比如某些消息不能马上 deliver,要等待 client 上线,怎么处理处理这些消息，database ? or queue?
  5. message queue 怎么存储?
  6. 传输信息丢失了怎么办，网络错误，是重发还是丢弃?
  7. ipv4 ipv6 互转
  8. message route for multicast peers
  9. 多语言支持
  10. 数据的编码管理

解决网络中复杂的连接与数据交换场景;

易用,轻量,快;

强调并行,跨线程,跨进程;

模糊掉 client 和 server 的概念，万物连接更像人的大脑，只有节点，没有绝对的中心;相比 http 为何不好?

it arguably makes the problem worse by encouraging developers and architects to think in terms of big servers and thin, stupid clients.

总的来说:

  1. It delivers blobs of data (messages) to nodes, quickly and efficiently.

  2. You can map nodes to threads, processes, or nodes. ZeroMQ gives your applications a single socket API to work with, no matter what the actual transport (like in-process, inter-process, TCP, or multicast).

  3. It automatically reconnects to peers as they come and go.

  4. It queues messages at both sender and receiver, as needed.

  5. It limits these queues to guard processes against running out of memory.

  6. It handles socket errors.

  7. It does all I/O in background threads.

  8. It uses lock-free techniques for talking between nodes, so there are never locks, waits, semaphores, or deadlocks.

从应用上看，消息中间件经常解决的问题:

  1. 网络通信(看成１个网络库?)
  2. 应用解耦
  3. 应用消息通讯
  4. 流量削峰( 消息缓存)
  5. 分布式消息通信

zeromq example

https://www.ibm.com/developerworks/library/se-distributed-apps-zeromq-part1/index.html

client 访问某个 url 从别的机器的 python machine learning 处理中心要结果，web server 是 node server, 如预测值.

client <--http--> node.js server  <--zeromq--> python

https://medium.com/@amit.kulkarni/consuming-services-for-web-using-zeromq-5d9817fec2cd