Algorithm of ESXi cluster migration between different vCenter servers with the ability to rollback

Abstract

Introduction: Hypervisor managers may become deprecated or outdated or may fail. In addition, customers may want to rebalance hypervisor load. All these can cause problems related to the incompatibility of software with different manager versions or to the level of application integration. An effective solution to such problems is to automatically migrate the configuration and resources between hypervisor manager instances. Purpose: To develop an algorithm of the automatic migration of configurations and resources of an ESXi cluster between different vCenter Server instances with the ability to rollback in the event of a failure. Results: An automatic migration algorithm has been proposed. On its basis a program module for enterprise storage has been developed, which allows migrating an ESXi cluster on which a storage system is deployed from one vCenter Server to another. The module handles errors, and in case an exception or error occurs the module initiates a rollback operation in order to revert to the original vCenter Server. The study presents an experimental evaluation based on the comparison of the manual migration and the execution of the developed module. The time needed for the migration procedure has been reduced from 55 minutes to about 3 minutes, i.e., by 15–18 times. The amount of information to be known or taken into consideration by a user has also been reduced. Practical relevance: The study results have been used for enterprise storage. The developed module makes migrations to a new vCenter Server easier and faster.