Large-scale server cluster application deployment method and large-scale server cluster

Abstract

The invention relates to an application and deployment method for a large-scale server cluster and a large-scale server cluster thereof. The application and deployment method comprises the following steps: step 1, a central deployment server in a large-scale server cluster divides an application deployment pack into a plurality of segments, and deploys the divided segments to a plurality of application servers in the large-scale server cluster; and step 2, after all segments of the deployment pack are deployed to all application servers, all segments are spliced into a completed deployment pack. The application and deployment method for a large-scale server cluster and the large-scale server cluster thereof relieve the performance load of the deployment server, improve the velocity and efficiency of distributing the application deployment pack to the application server to the deployment server, remove the processing and network resource bottleneck of the deployment server, eliminate the relationship between the time of application deployment and the quantity of application servers in the cluster, do not need to maintain a great quantity of layered deployment nodes as the C / S frame of the traditional layered distributing technology, and further lower the difficulty of development and maintenance.

Description

technical field [0001] The present application relates to application deployment of large-scale server clusters, and in particular to a method and cluster for realizing application deployment in a back-end server cluster of a large website (for example, a scale greater than 2000 servers). Background technique [0002] In today's Internet giants, such as Google, Facebook, Yahoo and other websites, in order to support the access of a large number of users, there are often tens of thousands of server clusters at the back end to provide services. The mature application deployment technology originally used in the operation and maintenance of a cluster with only a dozen servers exposed performance bottlenecks when maintaining tens of thousands of servers at the same time. [0003] For example, the R&D department has developed a new application. Whether this new application can be quickly and stably deployed to thousands of back-end servers and provide services to users in the fas...

AI Technical Summary

Problems solved by technology

Deployment nodes at each level must be managed, operated, and maintained independently, and each deployment node must develop different deployment codes for distributing application deployment packages to different application servers, which causes serious waste and makes servers The maintenance and management of the cluster is becoming more and more complex

[0009] 3. At the same time, the known deployment technology based on C / S architecture cannot meet the needs of rapid deployment of applications in large-scale server clusters. A cluster deployment of thousands of servers often takes 1 to 2 hours, seriously greatly affected the timeliness of application launch

Images