It is a distributed service framework, service registration and discovery, load balancing, SPC (remote procedure call).
Two, zookeeper middleware
Hadoop, which is a distributed big data processing platform designed based on Google's three papers on big data.
Paper 1: A distributed file system requires a large number of servers to store a large number of files in the field of big data (data above P level). All files must be safe. When the file data on a certain host is damaged, it cannot As for making this copy unavailable (one copy of data is stored in multiple places). In addition, the file cluster is managed by a host, which stores the addresses where all files are stored. In this kind of cluster structure system, if the host hangs up, the whole system will not work. In order to solve this problem, Apache has developed a sub-project in Hadoop, one of the top-level projects, which is the zookeeper project. Hadoop is an elephant, zookeeper is a zookeeper, and Hive is a bee. zookeeper assumes the role of distributed service manager in hadoop. The main data structure inside it is a tree structure.
Paper 2: MapReduce
Paper Three: HBase (Column Data Storage)
The main role of zookeeper
Although zookeeper was created based on hadoop projects, it is now used in many distributed systems for cluster management to achieve high availability of clusters. In addition, it can also manage distributed services, and can also implement distributed transactions and other functions. In addition, the function of distributed lock can be realized.
Dubbo's official recommendation is to use zookeeper as the registration center. All services in the distributed system must be registered in zookeeper.
install zookeeper
In the production environment, it must be installed as a zookeeper cluster, and the number must be an odd number of 3, 5, 7.... Developers can install it as a stand-alone machine, and operation and maintenance personnel should install it as a cluster.