Method for improving reliability of distributed system based on DDS technology

Abstract

The invention discloses a method for improving reliability of a distributed system based on a DDS technology. The method comprises the following steps: (1) determining a target application that a distributed system needs to perform redundancy processing, and decoupling the target application and an original information transmission relationship by using a publish-subscribe mechanism; (2) achieving independent running and redundant hot standby switching of the target applications by adopting a QoS strategy of the DDS technology, and ensuring that the process specifically includes the following steps: when multiple running instances that are redundant to one another of one target application are simultaneously run, only enabling the running instances with a higher priority to obtain the ownership of a subject; and when the running instances with the higher priority have failure, enabling the running instances with a secondary priority to obtain the ownership of the subject; and (3) after the running instances with the higher priority of the target application are recovered from the failure, enabling the running instances with the higher priority to obtain the ownership of the subject. The method disclosed by the invention has a lower resource overhead for providing a redundant hot standby switching function, is short in switching time, and is applicable to the distributed system of real-time tasks.

Description

technical field [0001] The invention relates to information system technology, in particular to a method for improving the reliability of a distributed system based on DDS technology. Background technique [0002] Distributed system is a widely used information system, which has the characteristics of many nodes, loose interconnection, separate development of software and hardware, etc. Distributed systems are generally composed of collaborative operating software deployed on multiple nodes. These software exchange information through the Internet to jointly complete a certain function or provide external services. During the operation of the distributed system, some nodes may have network link failures, switch failures, node hardware failures, node software failures, etc. The local failures of these nodes will cause the global functions of the distributed system to fail to operate normally, reducing the System availability. [0003] Since various hardware and software fai...

AI Technical Summary

Problems solved by technology

Adding a fault redundancy mechanism to the original distributed system that does not have high availability involves all-round modification of the original system from software architecture, to design, to implementation, with heavy workload, high cost, high risk, and long construction period

Especially when the software systems in the system are separately developed by different units, the implementation is more difficult

[0006] (2) The complexity of the software system increases, and the software system is difficult to upgrade and maintain

Using virtual machine snapshots to realize online backup and recovery of key components in a distributed system has a large network overhead, so that the switching time for providing redundant hot backup can only be at the second level, which cannot support real-time applications

[0009] (2) Unable to cope with the failure of the software itself

Especially when a software runs into an abnormal state, the virtualization management software still regards it as a normal running software for backup, and cannot perform redundant switching and fault recovery on it.

Images