Cloud deployment Combat: Framing and setting up a configuration database

When cloud computing has just been known, most people hear about public clouds. This is mainly because the public cloud service providers are some of the industry's big companies, such as Google, Amazon and so on. By the time people actually started to let cloud computing fall to the ground, the focus was on private clouds. This is because the data security, controllability and other aspects of consideration, the enterprise is more willing to choose private cloud. In view of this, it is necessary to understand how to deploy a private cloud.

In general, private cloud construction involves implementing virtualization, deploying automation tools, developing new processes, management systems, and so on, although some people think that automation and process orchestration tools are not the necessary cloud management tools, but the implementation staff and experts say Automated tools are as essential as the basic tools for managing servers and storage systems. Without these tools, you cannot reduce the cost of managing private clouds. In any case, selecting data Center virtualization software and selecting automation and orchestration management tools are critical.

How to get Started

How to set up a private cloud depends on what the business starts with, and the original environment may decide what you want to do first. If you're building a private cloud from scratch, you need to virtualize the server and then start virtualizing the storage and the network before you go to other areas.

If the enterprise wants to fully enjoy the benefits of the private cloud, these steps are necessary: you need to be able to configure the hardware and software they need for your users, and then deploy hardware or services; You also need a way to manage and control your IT operations environment; you need to be able to manage migrations from the private cloud to the public cloud, i.e. The ability to transfer workloads back and forth between the private cloud and the public cloud.

In general, the way to build a private cloud differs from enterprise to business, but there is always something in common. For example, when you are ready to build a private cloud, you should at least raise and answer questions such as which applications will run in the private cloud and which will not? Can you flexibly extend your application to make the most of the cloud? If you have two data centers, how do you migrate applications and share resources between the two? Which applications are deployed to the cloud are more advantages than disadvantages, which are more harmful than benefits?

These are questions that need to be thought over and over again, and businesses need to do their best to plan business processes for the private cloud.

Develop a framework and build a configuration database

One of the first things to do when Brisbane, a large financial services provider in Australia, Suncorp to plan a private cloud and develop a strategy is to create a set of service-based operating patterns and service catalogs. The service directory contains a series of automated services that are used internally, and are provided to internal users of the enterprise through Self-Service portal sites.

The key to creating a service catalog is to implement an information technology Infrastructure Library (ITIL) framework that stores information about corporate IT assets and business application relationships in the Configuration Management database (CMDB). All major IT processes in Suncorp: events, issues, assets, and changes are based on the CMDB.

The completion of the service catalog can be time-consuming and laborious, but if the enterprise is already using IT service management and change management tools, such as BMC Remedy or service-now.com, and has a ready-made CMDB, this work is more convenient. Enterprises can use the functions provided by the CMDB to manage the automation services listed in the service directory. Suncorp is doing it with the help of a CMDB based on the BMC Remedy.

Paul Cameron, Suncorp's Enterprise Services director, said Suncorp's deployment of private clouds was due to the need to provide better services to customers to meet their needs more quickly. In traditional data centers, businesses often take a week or even months to configure a server, and, of course, the length of time depends on how much work the IT staff is doing and how long the user teams are queued to process the tasks.

Now in Suncorp, users can request resources and services once they have entered the self-service portal. Once a request is made, these services are automatically delivered. Suncorp has virtualized most of its data center servers and storage systems, and about 80% of the data center's services are now available through automated self-service portals.

Companies with private clouds are most often using a way to control user needs, such as cost allocation, or a strict limit on the size of the user's capacity to request. Otherwise, it is entirely possible for users to constantly create virtual servers and quickly run out of existing capacity.

Capacity Planning

Consulting firm Precision It's system engineering division Jeffrey Driscoll that the most basic work that companies do when they start building a private cloud is to buy servers, storage systems (such as storage area networks, Sans) and virtualization software, and then start building clusters, followed by capacity planning.

Capacity planning involves doing the following: When you add servers and other resources to your cluster to meet your business needs, you need to be aware of the possible changes in requirements. Capacity planning is a major effort to ensure cluster and cloud performance. If the job is not done well, it will end up with a useless system, or have to install a traditional, non-cloud system to support business operations.

Most businesses are not good at monitoring and reserving computational power. In order to meet the needs of users, it is often necessary to provide more data centers, which means that a certain amount of hardware is idle. Documenting the use of your business in detail helps you ensure that you have enough computing power and storage capacity that is not too much. One solution is to create a hybrid cloud environment: when the computing power or storage capacity in the private cloud is insufficient, the user changes to the public cloud, such as the Amazon's Elastic Computing Cloud (EC2).

Once the cluster is running, you can start configuring the virtual server. The result is a layered architecture that includes the server layer, the network layer, and the virtualization layer. Each layer has a corresponding management tool, and this time you can start thinking about automation.

Assess Cloud security

Private clouds are thought to be good for businesses that are concerned about security issues or need to comply with regulations, but Suncorp's Cameron warns that private clouds force executives to reconsider how to treat security.

For example, the way to manage firewalls in traditional data centers does not always apply to cloud environments where workloads may be transferred at any time. The reason is that in a virtualized environment, the server may be divided into different security groups, and the security of the target host may not be able to meet the virtual machines that are migrated to it.

In a private cloud, virtualization firewalls are important. This is important because multiple virtual machines may be connected by using virtualized network switches and other virtualization components, unlike in physical environments, where the network is completely based on physical hardware and cabling.

One thing to keep in mind is that security issues in a virtual environment are not always exactly the same as security issues in a non-virtual environment.