Cloud computing is not available? The three major impacts of private cloud on it architecture

A private cloud is a cloud that is used by the enterprise itself, and all its services are not intended for others to use, but for their own internal personnel or branch offices. Private cloud deployments are more appropriate for large enterprises or government departments with many branches. Private cloud is deployed within the enterprise itself relative to the public cloud, so its data security, system availability can be controlled by itself. But its disadvantage is that the investment is bigger, especially the one-time construction investment is big.

As far as the current trend of cloud computing is concerned, the development of public cloud is obviously no private cloud fast, private cloud is the direction of big enterprise IT development. So how does a private cloud affect an enterprise's it architecture in the traditional it transition to a private cloud? This article focuses on the virtualization of servers and data centers to answer this question.

First: Problems with traditional data centres

Enterprise or organization in the past it process has followed such a process: business sector definition requirements-> Procurement-> System Delivery-> transfer to the data Center for operational management; Increasingly, more and more business units need to deploy their own IT systems and deliver them to data center operations and maintenance Management. Traditional data centers have the following main problems:

1. The low utilization of resources in data centers is a common problem in many data centers at present. Research by a number of third-party organizations shows that the average utilization of servers in the Enterprise data Center is less than 15%.

2. The application system construction is relatively independent and fragmented. From the perspective of infrastructure is an independent "island", it is difficult to consider the overall IT infrastructure in terms of resource allocation and use of rationality.

3, the traditional data center resource allocation and deployment process more manual, there is no corresponding platform support, so that a large number of human resources in the heavy duplication of work, there is no self service and automatic deployment capabilities.

Second: Enterprise traditional IT model and cloud computing model

The classification of cloud computing has many, more popular and recognized two kinds: according to the service pattern and deployment method to divide. Divide cloud computing According to service pattern: Software as a service (Software as a Service,saas), platform as service (Platform as a service, PaaS), infrastructure as service (infrastructure as a services, IaaS): The cloud can be divided into public cloud, private cloud, and community cloud, depending on how it is deployed. Enterprise traditional It model, private cloud computing model diagram as shown in Figure 1 below:

▲ Figure 1 Traditional IT and private cloud computing models

Some data centers currently include thousands of servers. As the number of servers continues to increase, we find that data center complexity and management costs are rising dramatically. Another problem facing large data centers is the problem of reliability. When the reliability of a service reaches 99.99%, this means that the failure time of one year is about 53 minutes. However, assuming that there are 10,000 such servers in the datacenter, maintaining high reliability can become a challenge. Although we can improve the reliability of a single server through redundant hardware design, this approach is costly and less effective as the number of servers increases. As a result, large data centers take the form of software to improve the reliability of the system, the hardware server through a software layer to form a logical computer. Such a server at the bottom of the failure can be isolated by the software, because the entire data center is large, so the overall computing power and reliability will not affect. The data centers of companies like Microsoft, Google, and Yahoo are typically built in this way.

Third: Dynamic Data center Solutions

On the IaaS of building a private cloud, Microsoft has proposed Dynamic Data center solution, is committed to helping enterprises or organizations to build secure and reliable, shared data center, for the users of the business system to provide the service mechanism, the realization of web-based IT resources internal leasing and accounting management, and for the high scalability of the system lay a solid foundation. Real business computing capacity and it asset stripping, the bottom of the data IT infrastructure to provide services, to meet the business variability and promote the rapid development of the business. The functions of Dynamic Data center include functions of self service, Service lifecycle management, resource management, resource service, security management and external interface. The products provided by the Dynamic Data Center solution are shown in Figure 2 below:

Figure 2 Ddtk-h and SC VMM SSP2.0

Microsoft Dynamic Data Center solution can help enterprises to create virtual environment to run applications, users can flexibly allocate appropriate application configuration as needed, and support dynamic expansion. Specific features include deployment, 24x7 monitoring, optimization, protection, and flexible adaptation in five areas. Among them, deployment capabilities include deploying resources such as servers, networks, and storage services, and flexible self-management. 24x7 monitoring capabilities include gathering operational data to better meet SLA needs, monitoring resource utilization, and monitoring customer self-monitoring. Optimization features include continuous monitoring and the ability to proactively adjust and migrate servers based on operational needs without impacting or impacting applications, and allocating "appropriate" resources as needed, with no matching and low matching. Protection features include anti-virus, garbage access filtering and firewalls, application and data backup, guaranteed 99.9% uptime and physical security of infrastructure. Flexible adaptation features include easy adjustment to the environment, deployment of new resources, storage, bandwidth, etc. that can be dynamically tuned as needed, support for different virtual technologies, and the ability to manage different types of virtual machines.

The logical architecture of the dynamic Datacenter solution is shown in Figure 3 below:

Figure 3 The logical architecture of a Dynamic Data center solution

The top tier is the service layer, which provides account management, service catalogs, deployment Services, and user reports; the next layer is the management layer, which provides resource management and load balancing; the next layer is the virtualization layer, which provides hardware virtualization and application virtualization; At the bottom is a resource tier that includes servers, networks, and storage. Ultimately help users implement the following features of the Dynamic Data center:

Resource pool Management: Centrally manage the hardware resources of the center, including server, storage, network, etc.

Dynamic Allocation Service: The platform can dynamically allocate service resources.

Self-Service Portal: Users can apply for computing resources on demand, and the platform determines the results of the approval based on SLA and user billing.

Application and Service Management: Application management, Service metrics billing, SLAs, data storage and disaster preparedness services.

Written in the last

"Each cloud service will have a different roadmap for future development--some cloud services should be focused on closer integration, customization and diversification of the business," said Tom Bittman, a senior analyst at Gartner. Others should focus on independence, simple interfaces, standardization, and customization. He added: "Many of the investments in private cloud computing can prepare companies to deploy public cloud computing." These investments are not just technical changes, but also changes in process, culture, and business interfaces. These changes should be decided sooner or later, this will help enterprises to better implement the decision of cloud procurement, but also very likely to enable enterprises to smoothly transition to the public cloud computing. "The future should be that with the virtualization of these large data centers, private cloud will become the dominant model for their IT systems."