Hot spot balancing method for metadata server

Abstract

The invention discloses a hot spot balancing method for a metadata server, which is applied in a metadata server cluster. The hot spot balancing method provided by the invention comprises the following steps: (1) dividing the metadata server cluster into a plurality of groups, each of which comprises a plurality of main metadata servers and a plurality of backup metadata servers; (2) constituting a hot spot shifting table on a client to record the information of the main metadata servers and the backup metadata servers when the client is mounted on the main metadata server; (3) monitoring the loading condition of each metadata server to judge whether the metadata server is a query hot spot; (4) and performing hot spot shifting on the corresponding metadata server, namely changing the hot spot shifting table and adjusting the hot spot shifting parameters in the table to lead the client to redistribute request load if the monitored metadata server is the hot spot. With the adoption of the hot spot balancing method provided by the invention, the hot spot problem in the metadata server cluster can be effectively solved, and the hot spot balancing for the whole metadata server cluster can be realized.

Description

Technical field [0001] The invention belongs to the technical field of computer storage, and particularly relates to a metadata server hot spot balancing method for a metadata server cluster in an object storage file system. Background technique [0002] With the rapid development of computer technology and the rapid progress of information globalization, traditional network storage methods can no longer meet today's application requirements. Generally, in a distributed storage architecture, file data and metadata are managed separately, and a dedicated metadata server is provided to manage metadata information. According to statistics, in the current file system, 50%-80% of the access is for metadata operations. On the other hand, although the storage space occupied by metadata is very small compared with file data, for mass storage systems, the storage space occupied by these metadata is also considerable, such as websites such as Facebook. The size of the catalog to be maint...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a metadata server hotspot equalization method, aiming at proposing an effective solution to the problem that the system service quality is severely degraded when a large number of accesses in a metadata server cluster are concentrated on a certain metadata server , and at the same time, this method also improves the reliability of the metadata server cluster to a certain extent

Images