Analysis of six kinds of architecture of cloud computing

Cloud computing requires good resiliency, scalability, automation, data mobility, multi-tenant, space efficiency, and support for virtualization. So what should be the architecture of the data center infrastructure in the cloud computing environment?

1. Cloud Computing Data Center overall architecture

Cloud computing architecture is divided into two parts: service and management. In the service aspect, mainly provides the user based on the cloud the various services primarily, a total of 3 levels: infrastructure, service IaaS, platform-service PAAs, Software-SaaS. In management, mainly cloud management, its function is to ensure that the entire cloud computing center can be safe and stable operation, And can be effectively managed.

2, Cloud computing room architecture

In order to meet the needs of cloud computing services flexibility, the cloud computing room uses a standardized, modular computer room design framework. The modular room includes the modular room of container and the modular room of building.

Container Modular computer room in the outside room without the application of the scene, reduce the construction side in the room location of pressure, to help the construction side of the original six months to shorten the construction cycle to two month, and energy consumption is only the traditional computer room 50%, can adapt to the desert hot and dry areas and polar cold areas of extreme harsh environment. Building modular room using cold air duct isolation, accurate air supply, outdoor cold and other leading refrigeration technology, can be applied to large and medium-sized data center building blocks and expansion.

3, Cloud computing Network system architecture

The overall structure planning of network system should adhere to the design idea of regionalization, hierarchy and modularization, so that the network level is clearer and the function is clearer. Data center Network According to the nature of business or network equipment for the role of regional division, can be from the following aspects of the planning.

1 According to the nature of the transport data business and user-oriented, the network system can be divided into internal core network, remote service network, public service network and other areas.

2 According to the different function of the network structure, the network system can be divided into the core layer, the convergence layer and the access layer.

3 from the Network Service data application business independence, the mutual exchange between the business and the security of the business needs to consider, the network system can logically be divided into storage area, application business area, front zone, System Admin area, hosting zone, external network access area, internal network access area, etc.

In addition, there is a fabric of the network architecture. With the deployment of cloud computing in the data center, the traditional network structure has the potential to make the network latency problem a major bottleneck, which makes it more urgent for low latency server communication and higher bidirectional bandwidth. This requires the network architecture to the flat direction of development, the ultimate goal is to minimize the number of network architectures between any two points.

One of the key to fabric network structure is to eliminate the concept of network hierarchy, Fabric network architecture can use array technology to flatten the network, the traditional three-tier structure can be compressed into two-layer, and eventually into a layer, by implementing the connection between any point to eliminate complexity and network latency. However, there is still no uniform standard for the new technology of fabric, and its application needs more practice.

4. Cloud Computing Host system architecture

The core of cloud computing is the concentration and scale breakthrough of computing power, and the computing type provided by the cloud computing Center determines the hardware infrastructure of the cloud computing center. From cloud customer demand, cloud computing centers often need to scale to provide the following types of computing power, its server system can adopt three (multi) layer architecture, one is high-performance, stable and reliable high-end computing, mainly dealing with tight coupling computing tasks, such calculations include not only external database, business intelligence data Mining and other key services, Also includes its own account, billing and other core systems, usually provided by enterprise-class large servers, and the other is for a large number of common applications of general-purpose computing, to provide low-cost computing solutions, this calculation for the lower hardware requirements, the general use of high-density, low-cost hyper-density integrated server, To effectively reduce the data center operating costs and end-user cost of use; third, for scientific computing, bioengineering and other services, to provide trillions, petajoules computing capacity of High-performance computing, its hardware base is high-performance cluster.

5. Cloud Computing Storage System architecture

Cloud computing uses a centralized data storage model, in the cloud computing platform, how the data is placed is a very important issue, in the actual use of the process, it is necessary to allocate data to multiple nodes of multiple disks. There are currently two ways to achieve this storage technology trend, one is to use a clustered file system similar to Googlefilesystem, the other is a block based storage Area Network San System.

GFS is a distributed file system designed and implemented by Google, which is based on a cluster system consisting of a large number of ordinary PCs with Linux operating systems, and the entire cluster system consists of a master and several chunkserver. There are several options available for San connectivity. One option is to use a fiber optic network that can operate fast fibre disks and is suitable for places with high performance and reliability requirements. Another option is to use Ethernet, iSCSI protocol, to run in a common LAN environment, thereby reducing costs. The use of SAN structure, the server to the shared storage device of a large number of data transmission through the SAN network, the LAN only assume the communication tasks between the servers, this division of labor so that storage equipment, servers and local area network resources are more efficient use of the storage system faster, scalability and reliability better.

6. Cloud Computing Application Platform architecture

The cloud computing application platform employs a service-oriented SOA approach that provides the infrastructure for deploying and running application systems, so application developers need not be concerned with the underlying hardware and application infrastructure of the application, and can dynamically expand the resources required for application systems based on application requirements.

(Responsible editor: The good of the Legacy)