Extreme learning machine-based slow disk detection method and system

Abstract

The invention provides an extreme learning machine-based slow disk detection method and system. By performing feature extraction on historical disk data, eigenvectors are selected from features to perform training, so that a scheme of detecting a slow disk based on a neural network is realized, a slow disk detection process is optimized, the accuracy of slow disk detection is improved, and the calculation complexity is lowered; and meanwhile, in an actual usage process, with increment of a historical disk data quantity, more and more samples are continuously trained, and model precision is higher and higher, so that the accuracy is further improved and a data storage system is ensured to be always in an optimal working state.

Description

technical field [0001] The invention relates to the field of computer storage, in particular to a slow disk detection method and system based on an extreme learning machine. Background technique [0002] With the continuous use of large-scale computer information systems, data storage systems such as cloud storage systems or disk array systems (RAID, Redundant Array of Independent Disks) have been more and more widely used. Data storage systems usually need to use a large number of disks and other storage media to store data. In the present invention, devices or devices that use at least one storage medium to store data are collectively referred to as a data storage system. [0003] As a data storage method, the function of RAID is to provide a storage system with super large capacity, fast response speed and high reliability in the form of a disk array when multiple disks are connected to a dedicated server. As another data storage method, cloud storage is a new concept e...

AI Technical Summary

Problems solved by technology

However, the above detection method does not consider the real-time conditions such as the disk workload. For example, when the disk is running at full capacity, if the processor continues to send I / O to the disk, the disk cannot respond to the subsequent I / O in time, resulting in subsequent I / O. / O is counted as a large response time. In these cases, it is likely to cause member disks of the data storage system to be misjudged as slow disks, resulting in normal disks being mistakenly judged as slow disks.

And there is no scheme for detecting slow disk based on neural network in the prior art

Images