Big Data Day "fifth: Distributed Coordination System" notes-how to get elected boss? How do I join an organization? How to pass the wishes of the organization

Chubby:

Maintain high availability and reliability, do not pursue the high throughput of read and write, all read and write requests have the master server complete, the other backup server in memory maintenance and master server fully consistent tree structure, the role of the subordinate server is the overall system reliability.

Zookeeper:

Each node can handle read requests and write requests can only be handled by the master server, where the role of the secondary server is to increase the system's read throughput (which may cause the data to be read as old data).

Zookeeper main application scenarios:

Election boss

Configuration management: Configuration files are stored on a node, other nodes are observers, Qidong can be read or subscribed to configuration messages.

Group member Management: Dynamically monitor changes in the membership of a team.

Task Assignment

Lock Management

Zookeeper for federated use with other architectures:

STORM: The storage place for the master process and worker process state information as a switchover between the master server and the backup server

MESOS: Leader election and primary backup server start switch

KAFKA: Automatically discovers and adds consumer servers (brokers) and consumer service providers (Consumer), keeping consumers and message queues mapped and consuming information in the location of message queues.

Katta: Primary backup server switchover and work state configuration information management.

Big Data Day "fifth: Distributed Coordination System" notes-how to get elected boss? How do I join an organization? How to pass the wishes of the organization