The infrastructure of cloud computing

Cloud computing Infrastructure

Based on the cloud computing scenario model we presented above, this section presents a basic architecture of cloud computing to provide a reference for readers. The framework for this architecture is shown in Figure 9.1.

In this framework we have to solve the following problems.

(1) The user is quarantined when the unreliable node fails.

(2) Computing and storage integration to accommodate computationally intensive and data-intensive tasks.

(3) Computing the virtualization of resources and storage resources.

(4) Isolation of data storage and computation between different applications and different users of the application layer.

(5) Application layer interface and the provision of operating system.

We have emphasized that this cloud computing system is designed based on untrusted nodes, the bottom of the framework is composed of a large number of such nodes, each server using the existing mainstream operating system, through various networks to connect all the nodes to become a huge cluster system, this part of the cloud of the physical hardware core, This core of complexity will be shielded by all users through the core layer of cloud computing software. These complexities include node failure, network failure, scheduling between nodes, load balancing, data security, data storage location, computing location, specific configuration of each node, and implementation of high availability of the system. Google's system gives us a good demonstration, we search in the search box while there are a large number of servers working for us, but all the details of the server's work we do not have to know, we are facing only a concise search box.

The computing and virtualization layers in this framework provide flexible computing and storage resources for computing resources and storage resource virtualization. The Distributed File system provides a file storage service with a replica policy for the system, and the security level of the file storage can be divided into: single level, cross server level, cross cabinet level and cross zone level. This file system not only provides a secure storage platform for the system, moreover, because of its distributed and replica strategy, it can provide storage support for computing layer, realize the migration of computing to storage, reduce the time of data transfer between networks, and realize the migration of computing and storage to backup nodes when the nodes fail. This enables high availability of the system in the event that the node is not trusted. Virtualization layer, file system layer and compute layer compose the software core layer of cloud computing infrastructure, it is also the technology layer that we need to conquer at the present stage, the storage, calculation and invalidation of cloud computing in the software core layer will be completely isolated, the application layer will not "see" the working details of hardware layer,

The application layer interfaces with cloud computing infrastructure through cloud operating systems and APIs, because it is located on the hardware core layer and software core layer, different users and applications in the use of the platform can only "see" is a "flexible supercomputer", the underlying software and hardware running details can not "see".

This framework enables cloud computing to hide the computing and storage resources of all nodes in the cloud, reflecting the beauty of cloud computing's "elephant Invisible" philosophy of technology.