dubbo mq and usage scenarios

MQ: message queue. Producer-consumer model, can be used for real-time news on the scene less demanding. Indirect call the relationship between multiple processes

Dubbo: RPC implementation. Direct call relationship between multiple processes

 

dubbo

1, rpc distributed cluster support: load balancing is provided outside a public address to share request came through polling, random form of pressure, hang one up a
2, combined zookeeper decoupling :( provider registration and consumers subscribe) client and server boot time will put their machines to the IP registration zookeeper. Ip server on the client will zk pull on the disk, and what ip record which services are available (the server to start when exposed to zk).
   Then call when the client calls the service will be based on ip end of service, even if this time zk hang it does not matter.
3: Communication connection length: NiO Communications Abstract packaging (being not in contact)

Available scene:
1, mall traffic soared to do activities: to prevent bounced off the system can be controlled by traffic Dubbo
2, distributed server procedure call rpc share pressure

 

mq a point a subscription is distributed

2 mq benefit is
1, the message is not lost: take away messages between service saved to mq middleware, server recovery when consumers will be re-made in the past, the message will not be lost
2, asynchronous processing: for example, a mall after users purchase will go back and update the database in response to the client, if in the case of high concurrency,
so the client will update the database in response to slow, consumers can use the process mq message queue to get data asynchronously write data Since confront the message processing service much faster than the database,
so that the response delay can be effectively improved

Guess you like

Origin www.cnblogs.com/lukelook/p/11229489.html