How to carry out server consolidation in cloud Era

[Guide] Server consolidation has been a number of successful application cases. As the pace of cloud computing and virtualization accelerates, more customers who have not yet done server consolidation are beginning to focus on this http://www.aliyun.com/zixun/aggregation/13748.html "> Infrastructure solution."

In China, server consolidation has many successful application cases. As the pace of cloud computing and virtualization accelerates, more customers who have not yet done server consolidation are increasingly starting to focus on this infrastructure solution.

Like many enterprises in the world's IT departments, the domestic IT application has also experienced a period of rapid expansion of server hardware equipment. Especially in 2005 years or so, with the unit calculation price greatly reduced, as well as server suppliers of homogeneous price competition, making the server purchase price becomes extremely low. And because of the financial model of many domestic enterprises, it is very common to establish a separate system for each application of each department. From 2007 onwards, we have found that it hardware expenditure in the overall it cost of the proportion of the declining, and room resources, energy consumption and other hidden costs are rapidly increasing.

Computing resources can bring advantages is self-evident, in addition to eliminate the hidden costs above, for domestic users, it is very important to face the changing business environment more quickly. The original system construction mode is generally internal budget application, project, procurement, arrival, installation, testing, on-line ... Usually lasts 3 months or more. And how to achieve full resource pooling, easier to monitor, forecast and deploy the whole system resources, and reduce the computational resources of those secondary applications, release more processing power. At the same time, if the peak resource demand for each server does not occur at the same time, it can also be the wrong peak integration. For example, some financial systems typically only have large transactions at the end of the month, and can reduce the response speed of other core applications.

In the server virtualization mode, there are both horizontal and vertical modes. We are now focusing more on Scale-out mode, integrating a large number of small servers into one computing pool, and on-demand resources. In fact, the vertical expansion of the way in the country also has a lot of applications. Here the pros and cons of the two and the application of the scene briefly introduced.

The most common way to expand horizontally is to build a virtual layer on an existing hardware platform and run the operating system and applications on it through virtualization software such as VMware. This way can make full use of the user's existing system resources, in some cases, through the corresponding system analysis software (such as VMWAREROITCO, etc.) analysis of integration, without the need to purchase any additional server equipment, you can obtain a larger computing resources.

But this is not a panacea, some domestic customers often have such a scenario, that is, using the original 2 servers (each running 2 processors) to run a 4-processor-required application, the idea is not supported in the horizontal virtualization way. (High Performance Computing application exception) In addition, this kind of virtualization integration will also bring some hidden costs, such as the integration after the failure of how to quickly find suppliers to provide service response, many domestic x86 architecture of the server general use of the cycle is 5 years, Over a 3-year warranty period, users will need to pay for the maintenance costs of the equipment vendors, plus the consulting, deployment and maintenance costs of additional virtualization software, and the overall cost of ownership scaling is not necessarily lower than the vertically scalable virtualization consolidation approach.

Furthermore, the way of scaling up requires a large-scale migration of the current application environment. This is not an easy job for users to do.

Vertical scaling refers to migrating applications from multiple devices to a high-end device, such as consolidating 4 2-way servers into 1 8-way servers. The biggest benefit of this approach is that it improves manageability so that when a failure occurs, the location of the fault occurs more quickly, and the overall number of servers is reduced. and vertical scaling can be a more secure way to migrate existing applications to new high-end platforms without any resource constraints caused by planning errors.

The disadvantage of vertical scaling is the large servers that require additional purchases, which is a cost to the enterprise, especially when there is a large number of servers to consolidate.

A better approach is for users to fully evaluate an existing application (rather than simply running the resource analysis software). The application is divided into two categories, key core applications such as finance, billing, customer relationship systems, and non-critical applications such as mail, web portals, etc. (for different types of enterprise classifications). Deploy core applications on separate servers to ensure performance and consider deploying a clustered environment. And the Non-core can also be used to run in the use of a long time calculation pool, through the virtualization software fault tolerance to ensure that the system will not downtime. In this process, the use of existing equipment, fault conditions and server vendor Maintenance period is also a factor to consider.

(Responsible editor: admin)