The Research On The Automation Deployment Of Operation Systems

The IT infrastructure of large enterprise often has a very large-scale servers, the totalnumber of servers can reach as much as thousands. It is very challenging to deal with thematters of deploying, upgrading and reinstalling the system. In the meanwhile, the server’sdeployment configuration, running platforms and services also differ in different sectors. Onthe other hand, the very popular virtual machine deployment technology recently willgreatly increase the number of virtual servers, the frequency and difficulty ofthe installation process, which has put forward higher requirements for the deployment.Automatic deployment must be taken into consideration after large scaling for eachcompany’s IT infrastructure, which can greatly liberate the burden of administrator, improvethe efficiency and reduce error rates.However, there still existes many problems during the implementation of the automaticdeployment. On the one hand, many companies have not yet realized the importance andcomplexity of automated deployment, so they will tend to choose a relatively simple butinextensible scripts operation method. On the other hand, most of the automated deploymentsoftwares have a limited scope and exist various problems, which cannot meet the individualneed of different companies and is not unable to form a unified automated deployment idea.In this paper, the IT resource deployment in a large online-game company is based to tryto select the most appropriate automatic deployment platform considering the features thatmultiple operation systems, high security levels and the support for “puppet”, themanagement tool, are needed. Some significant features of this system can be shown from theresults of the tests:1. The template-based system installation mode translates the need of compatibility withdifferent operating systems into that of the different modules. Developers have the flexibilityto develop a variety of customizable templates based on existing templates to meet the needsof deployment of most versions of the operating system.2. The SVM-based state transition mechanism can store all the status during the wholeinstallation process, and can transfer the process to a different state based on operations of thenodes. And all the installation process is determined according to this state. This mechanismcan, on one hand, prevent erroneous installation conditions. And on the other hand it can offergreat convenience to the custom installation process3. The deployment of the high-performance Nodejs as the mirror provider can greatly ease the bottlenecks that constrain the performance in the installation process, which is thedownload speed of resources and softwares.4. The system is fully fused with some system configuration and management softwares,such as puppet and chef, which can eliminate the need to manually deploy such neccessarysoftwares, so as to seamlessly be integrated with the following management system.