Hotspot data identification method and device

Abstract

The invention discloses a hotspot data identification method and a hotspot data identification device. The method comprises the steps of determining an initial hotspot queue according to the visit times of each storage data block within a first preset time slot; computing the weight of each hotspot data block in the all hotspot data blocks and the association degree between each data block in a data block set associated with the hotspot data blocks and the hotspot data blocks at preset intervals; multiplying the weight of each hotspot data block by the association degree between the hotspot data block and each associated data block, thus acquiring a heat prediction value of each data block; sorting the all data blocks with the heat prediction values according to the heat prediction value, wherein the data block with the plurality of heat prediction value is subject to the own highest heat prediction value; and adding the former preset quantity of data blocks in a sorting result into the initial hotspot queue to acquire the current hotspot queue. According to the method and the device provided by the invention, the hotspot data can be effectively identified, and the situation that the data is upgraded, degraded and migrated repeatedly in short time due to sudden visits can be avoided as much as possible.

Description

technical field [0001] The invention relates to the technical field of hotspot data management, in particular to a hotspot data identification method and a device thereof. Background technique [0002] In order to improve I / O performance, multi-level storage media are currently used to store data with different access temperatures. A large amount of cold data that is not commonly used is stored in a storage medium with slow read and write speed and low cost. Hot data is stored in storage media with fast read and write speeds, which can not only reduce storage costs, but also improve I / O performance. [0003] The existing hotspot data identification method mainly divides the storage area into several data blocks, and then counts the access times of the data blocks, counts the access times of all data blocks in a certain period of time, calculates the access frequency according to the access times, and The data with high access frequency is put into the hotspot queue as hotsp...

AI Technical Summary

Problems solved by technology

[0004] However, when there is a sudden access demand, there will be a large number of accesses in a short period of time. In this case, the above identification method will mark the data block with sudden access as hot data and put it in the hot queue. , the data block may only be accessed once and then will not be accessed again. Therefore, in this case, data will be repeatedly degraded and migrated in a short period of time, which will increase the burden on the storage system and cause data jitter.

Images