In the era of big data, the transition from client/server to node in computational mode

In the database age, the role of computers in the distribution system is clearly divided, not the client is the server, usually a server attached to multiple clients, the server assumes the storage and computing work, the client is responsible for displaying the server processing results. High-performance computers, such as small opportunities, are made as servers, and low-end computers, such as personal computers, become clients. This is the Client/server (client/server) structure that you used to say.

In the age of big data, this role has changed quietly. The concept of client/server has been blurred and replaced by the concept of "node". The reason for this change, in the final analysis, is that the data processing demand has undergone an essential change, forcing the data processing model to correspond to changes. An experienced data system administrator might as well compare how much data is stored and calculated in the database, and how much of that data is stored and calculated in a big data environment. It is because the data storage and computation of Big data cluster is too big, if continue to process the data according to the C/S mode, can not adapt to this huge data processing requirements, but also the cost of equipment procurement and operating costs of pressure, there is an urgent need to change the system operation mode, redesign system architecture. In this way, the concept of nodes emerges. In today's big Data Environment, data processing is designed to be divided into functions, cut into different stages and modules, each node simultaneously assume the dual role of client and server, that is, send instructions and processing instructions. When sending instructions, it is a client that waits for feedback from the server, and when the instruction is received and processed, it is a server that performs the data processing work. This change in roles and working patterns, both simplifying the benefits of data processing design and processes, also makes us less concerned about the demands of computer performance. In today's big Data cluster, the PC is able to completely substitute for the minicomputer, which is why. And because of the need to further reduce costs and save energy, mobile architecture hardware, represented by mobile phones, is increasingly entering the big Data environment, which is beneficial to accelerating the popularization of big data.

This article is from the "11187295" blog, please be sure to keep this source http://11197295.blog.51cto.com/11187295/1741652

In the era of big data, the transition from client/server to node in computational mode