Realization method of distributed consistency of data screening rule system

Abstract

The invention provides a realization method of distributed consistency of a data screening rule system. The method avoids the situation that the system is globally unavailable due to abnormity of nodes in the centers of rules in a centralized manner; and the method can ensure that all the established and effective rules are examined and approved. The method adopts the solution of a centralized and distributed double-stack framework in which two stacks simultaneously operate and do not interfere with each other on a wide area network level, and inconsistent states of results in the two stacks are reconciled at a terminal, so that a centralized examination and approval function and partial availability priority are realized; auditing is used to replace examination and approval as a concession condition to support partial availability priority constraint.

Description

technical field [0001] The invention relates to a distributed consistency realization method of a data screening rule system, which belongs to the field of modern data filtering. Background technique [0002] The data filtering rule system will be applied in a multi-user, distributed and wide-area network environment. Its data center is often at the end of the network, with large resources and low interconnection bandwidth. There are also certain constraint strategies between resources and users. Users can directly manage local resources, while remote users need to connect to the resource end through comprehensive utilization of the server. [0003] Different from the "decentralization" feature of general distributed systems, the data screening rule system requires a control / arbitration role to complete the centralized approval of the rules to achieve the screening of the legality of the rules. This requires us to design a centralized control scheme. Distributed wide-area ...

AI Technical Summary

Problems solved by technology

[0011] 5. In the Paxos algorithm, the proposer, acceptor, and learner cannot promise that others will appear when they need it, nor can they promise the time for approving resolutions or delivering messages

[0013] 1. The messages in the rule system may be lost during the transmission (store and forward) process, while the Paxos algorithm requires that the messages must arrive reliably

[0014] 2. The rule system requires that in the event of a WAN failure, the resource user can still manage the resource device, but the Paxos algorithm cannot work normally when more than half of the nodes fail, and the resource user cannot manage the resource device

[0015] 3. The rule system requires the rule center to approve the rules, and the Paxos algorithm has the characteristics of decentralization, and the nodes are completely equal, without any central node, so the approval function cannot be performed

Images