Data center network structure with high fault tolerance and high extensibility

Abstract

The invention discloses a data center network structure with high fault tolerance and high extensibility. A structural unit is formed by four switches, wherein three switches are mutually connected toform an equilateral triangle with a side length of l, and the other switch is located at the central position of the equilateral triangle and is mutually connected with the other three switches; theabove structural units are tightly connected to form an N-grade single-layer structure, and an equilateral triangle with a side length of Nl is formed after removal of repeated nodes, wherein N is aninteger greater than or equal to 1; m single-layer structures are arranged in a layered manner to form the data center network structure, wherein m is an integer greater than or equal to 1; the distances between each two adjacent layers are the same; and the corresponding switches between the layers are connected, and each switch is connected with a terminal node; one switch is connected with a client in each layer, and other switches are connected with a server; and the switches connected with the clients in adjacent layers do not correspond to each other. The data center network structure isadvantaged by high data transmission rate, high fault tolerance, and good load balance.

Description

technical field [0001] The invention relates to a data center network structure, in particular to a data center network structure with high fault tolerance and strong expansibility. Background technique [0002] With the rapid development of cloud computing, Internet of Things and other emerging Internet services, the world has entered the era of data explosion; according to the Digital Universe report released by International Data Corporation, it is estimated that the total amount of global data will reach an astonishing 40ZB by 2020; Cisco In "Internet of Things Products and Application Cases Introduction and Innovative Product Release", it is proposed that by 2020, there will be an increase of 50 billion connected devices in the world, and these connected devices will generate a large amount of data; as the core infrastructure of new application services such as cloud computing , the data center has become the focus of academia and industry in recent years; the surge in ...

AI Technical Summary

Problems solved by technology

[0003] Due to the problems of low transmission efficiency, limited cable bandwidth, and large transmission energy consumption in the traditional electrical interconnection tree data center network based on the binary on-off keying OOK modulation format, with the continuous growth of information flow, the throughput, energy efficiency and There are serious development bottlenecks in security and other aspects; for the big data processing requirements of cloud computing, the traditional data center network is far from being able to play a good role; because of the characteristics of the tree topology, the traditional data center network is In terms of reliability, there is a problem of network single point failure. Once the network scale is large, the failure of high-level equipment will cause some lower-layer links to be unusable; the classic Fat-Tree network structure eliminates the throughput limitation of the traditional tree-structured upper-layer links. However, its ability to deal with switches is insufficient, and it is sensitive to the failure of the underlying switching equipment, so it still has limitations in fault tolerance; the existing topological solutions that introduce optical switching, such as VL2, OSA, Helios, etc., rely on optical switching networks for upper-layer applications The device provides high bandwidth and dynamic configuration of network resources to effectively reduce the complexity of networking, but inevitably there is a delay problem and limits burst traffic, thus reducing its fault tolerance

Images