Online fragmentation capacity expansion method and device for a distributed system cluster

Abstract

The invention provides an online fragmentation capacity expansion method and device for a distributed system cluster, which can be used in the financial field or other fields. The method comprises the following steps: updating a cluster routing configuration rule in a distributed system according to a preset maximum cluster fragment number; determining a to-be-fragmented cluster in a plurality of clusters of the distributed system according to the updated cluster routing configuration rule, and splitting the to-be-fragmented cluster to obtain a new fragmented cluster; according to a received external data request, determining a Hash value corresponding to the external data request by using a Hash algorithm, and calculating a remainder value of the Hash value and the maximum fragment number of the cluster; and comparing the remainder value with the updated cluster routing configuration rule, determining a specified cluster corresponding to the remainder value, and routing the cluster data corresponding to the remainder value to the specified cluster. According to the invention, fragmentation capacity expansion under the distributed system is realized, and the hotspot cluster is split and expanded, so that the expansion nodes, the change time and the influence on a service system are reduced, and the resource cost is saved.

Description

technical field [0001] The invention relates to the technical field of distributed systems, in particular to a distributed system cluster online fragmentation expansion method and device. Background technique [0002] At present, when a new business system is first launched, the business volume is small. In order to save resources and costs, the smallest cluster deployment is generally adopted. For systems using unitized or microservice architectures, the database and application containers are deployed together. Database sharding generally adopts the method of moduloing the initial cluster after the hash of the business field to determine the cluster where the data resides. The formula is expressed as: mod(hash(business field), number of clusters n). The number of clusters n is relatively small at the initial stage of business launch. Under the deployment architecture of active-active in the same city, n generally takes an even number, and the two parks are deployed equall...

AI Technical Summary

Problems solved by technology

However, in actual production, the distribution of data may not be uniform. Some clusters are relatively active in business and the amount of data increases rapidly. The performance and capacity of the database of these clusters will soon be insufficient.

Doubling the capacity every time not only wastes resources, but also involves full data migration every time, which has a large impact and a large workload

Images