A cold data placement strategy based on frequency association in a low-energy cluster environment

Abstract

The invention discloses a cold data placement strategy based on frequency correlation under a low-power-consumption clustering environment. According to the technical aim, part of storage nods of a large number are kept under a low-power-consumption state for a long time by existing energy-saving network clustering storage systems and are started only when need to provide service, so that the energy saving aim is achieved, and with the help of data access frequency and relatedness, the related cold data is centralized to a same low-power-consumption node. Accordingly, multiple pieces of related cold data are accessed by using the strategy, cold nodes only need to be started once under an ideal situation, and the starting number of the cold nodes is reduced; the related data is usually accessed together, accordingly, the calling number of the nodes can be effectively reduced through the strategy, and meanwhile response time delaying to cold data access is reduced. In addition, the related cold data is written into a disk by using the strategy constantly, reading and writing access delaying can be reduced, and the system performance is improved.

Description

technical field [0001] The present invention relates to the field of computer storage technology, in particular to a cold data placement strategy based on frequency association in a low-energy cluster environment, and more specifically to a method for classifying data based on data access conditions and identifying data access associations, and The associated cold data is dynamically migrated to the same cold node. Background technique [0002] With the popularity and rapid development of the Internet, the number of Internet users has become difficult to estimate. The network has covered all aspects of human production and life. In daily production and life, Internet users will generate various data and information, such as learning materials, office materials, business data, game entertainment data and programs, etc. This makes the network storage system need to store and process huge data. Therefore, the scale of network storage systems is increasing day by day, so its ...

AI Technical Summary

Problems solved by technology

However, because waking up a closed (or dormant) node consumes a lot of power and will cause a certain delay, when the node evokes too many operations, instead of achieving energy-saving effects, it will increase energy consumption and degrade system performance.

Images