The Design Of System Management Software For Autonomic-computing-oriented High-availability Server

Application of computer servers in special areas such as banking, e-commerce,telecommunications, places much importance on their performance and availability.This type of server requires a high availability design with distributed architecture.Based on the background of the subject, this paper investigates the design ofmanagement software for servers of high availability and with distributed structure.This software promises to automatically manipulate the running of the serversystem, by ensuring components carry out their duties, cooperate, and orderlyoperation.This paper first introduces the source and background of the subject, describesthe overview of the development of related fields, including a technical overview ofautonomic computing, high-availability technology profiles and high-availabilityserver management software research. After that, the architecture of the targetsystem is described, and the role and status in the system of the main componentsare described. On the basis of comprehensive and detailed system requirementsanalysis, this paper presents a summary of the management software’s design.Finally, the article elaborates the detailed design of the system managementsoftware, including a specific description of the architecture design, detailed designof major components and core algorithms.This article focuses on the architecture of the system management software,the task of the distributed structure of a highly available system for load balancingscheduling policy design and the design of the virtual network managementsolutions, distributed high-availability systems availability management programsof the two-machine system design. Software architecture design, through the sharedobject for Linux technology, C++object to dynamically create the use oftechnology and a specially designed interface class, so that software can be rundynamically load, unload and replace function module. This makes the softwarewith the online upgrade, the line extended functionality features. On the task of loadbalancing, scheduling, design four kinds of common scheduling policies andpropose a new a new strategy based on dynamic feedback mechanism. Whilekeeping good load balancing, this strategy can reduce the time complexity of the algorithm and improve the scheduling efficiency of the implementation. For virtualnetwork management, this paper implements a incremental connection backupmechanism on the basis of a core network address translation module, in order toachieve a quick backup of the virtual network connection, which makes continuityof service after the Leader failover possible. This article is based on active faultmanagement system to make high availability for the two-machine sub-systemdesign. Leader node failure is detected by the fault management system and notifiedto management software to switch the two Leaders.