Data center in a cloud computing environment, network and system architecture
For purposes of the cloud, the cloud from the art should focus on improving high-end server infrastructure, low-cost high-density servers, mass storage devices and other computing devices and high-performance computing data processing capacity of the data center. Cloud computing requires infrastructure has good flexibility, scalability, automation, data movement, multi-tenancy, space efficiency and support for virtualization. So cloud computing architecture of each part of the data center infrastructure under what circumstances should it?
1, the overall architecture of cloud computing data center
Cloud computing architecture and service management is divided into two parts. In services, mainly to provide a variety of services based on a user's cloud-based, contains three layers: infrastructure as a service IaaS, Platform as a Service PaaS, software as a service SaaS. On the management side, mainly in the cloud-based management, its function is to ensure that the entire cloud computing center can secure, stable operation, and can be effectively managed.
2, cloud computer room infrastructure
To meet the needs of cloud computing services flexibility, cloud computer room use of standardized, modular room design architecture. The modular engine room and engine room including container modular building modular room. Modular containers in outdoor room without room scene application, reducing the pressure on the construction side in terms of the location of the room, and half of the original construction period shortened to two months, while energy consumption is only 50% of the traditional room can accommodate hot, dry desert areas and extreme environments polar cold areas. The modular building using hot and cold air isolation room, precision air, outdoor cold source other leading refrigeration technology, applicable to modular construction and expansion of the medium and large data centers.
3, cloud computing network system architecture
Network system architecture planning should adhere to regionalization, hierarchical, modular design concept, the network hierarchy clearer, function more clearly. Area division data center network action according to the nature of traffic or network device, the content can be planned from the following aspects.
1) Nature and transmitting data traffic for different users, the network system can be divided into the inner core network, remote business network, like the public network service area.
2) according to different network structures acting devices, the network system can be divided into core layer, convergence layer and access layer.
3) independence of data from applications business network services, safety isolation requirements of each business relations and mutual visits of business into account, the network system can be logically divided into storage areas, application service area, front area, the system management area , escrow, Extranet access zone, the internal network access zones.
After the deployment of cloud computing in the data center, traditional network architecture has the potential to become a major problem of network latency bottleneck, which makes low latency between the server communications and more two-way bandwidth requirements become more urgent. This requires the development of the network architecture flat direction, the ultimate goal is to minimize the number of network architecture between any two points, in order to eliminate the complexity and network delay.
4, cloud computing host system architecture
Cloud computing is core computing power of concentration and scale of a breakthrough, the type of cloud computing centers provide external determine the hardware infrastructure cloud computing center. From the cloud customer demand, cloud computing centers typically need to provide large-scale computing power following types of its server system may employ three (multi) layer architecture, one high-performance, stable and reliable high-end computing, mainly deal with tight coupled computing tasks, such calculations include not only the external database, business intelligence, data mining and other critical services, including their own accounts, billing and other core systems, typically provided by large-scale enterprise-class server; the second is for many ordinary general-purpose computing applications , for providing a low cost computing solutions, this calculation requires less hardware, typically high-density, low-cost ultra-dense integration server, in order to effectively reduce the cost and operating costs of the end user of the data center; is oriented three scientific computing, bio-engineering and other services to provide one hundred trillion petaflop high-performance computing power of computing, the hardware is based on high-performance clusters.
5, the cloud storage system architecture
Cloud computing uses data centralized storage model, the cloud computing platform, how the data is placed in a very important issue, in actual use, the need to distribute data to multiple disks among multiple nodes. And to achieve this goal of current trends in storage technology, there are two ways, one is similar to the Google File System using a cluster file system, the other one is the storage area network SAN-based systems block device.
GFS is designed by Google Inc. and implemented a distributed file system, cluster system based on a large number of ordinary PC equipped with the Linux operating system configured. On the SAN connection, there are many options. One option is to use fiber-optic network, capable of operating fast optical disks, suitable for high performance and reliability requirements of places. Another option is to use Ethernet, take ISCSI protocol, capable of operating in a normal LAN environment, thereby reducing costs. SAN fabric uses the server to the shared storage volume data transmission is performed through the SAN network, a local area network communication tasks bear only between servers, storage devices such that this division, a LAN server, and more efficient use of resources, the memory speed of the system faster, better scalability and reliability.
6, cloud computing application platform architecture
Cloud Application Platform by way of service-oriented architecture SOA and application platform to provide the necessary infrastructure resources application infrastructure to deploy and run applications, so application developers do not care about the underlying hardware and application infrastructure applications, and depending on the application dynamic resource requirements needed to extend applications. A complete application platform architecture provides the following functions.
1) application execution environment: a distributed operating environment, a plurality of types of data storage, dynamic resource stretching;
2) Application Life Cycle Support: Provides development SDK, developers IDE and so speed up the application, testing and deployment; public services, such as providing public services API forms, such as queue services, storage services and caching services; and monitoring, management and metering, pool resources, management and monitoring application system, accurate measurement. Applications using computing resources consumed.
3) integration, the ability to build composite applications: In addition to providing the application execution environment, but also need to provide connectivity services, integrated services, messaging services, services and processes used to build composite applications SOA architectural style.
These are some of the analysis computing data center architecture to the cloud. Cloud computing is called "cloud" because it has a realistic cloud characteristics in some respects: the cloud are generally larger; scale cloud can dynamically stretch, its boundaries are blurred. Cloud computing business model to provide users with an IT service, its content also changes over time, the dynamic elasticity. Therefore, cloud computing data center architecture with social progress will continue to adjust and optimize.