IO processing method, device and terminal in storage system

Abstract

The invention discloses an IO processing method in a storage system, which comprises the following steps: merging data of overlapping IO in a chain list to form merged data; Performing a down brush operation on the merged data; After brushing the merged data, the corresponding response function is called to complete the IO processing. The method fully utilizes the existing linked list of overlapping IO, merges the data of overlapping IO in the linked list first, and then performs down-brush operation on the merged data, that is to say, merges the data of overlapping IO in advance, so that themerged data can be brushed down once and returned in batch; Further, the number of reads and writes to the back-end disk can be reduced, the delay of IO processing can be reduced, and the IOPS of thestorage device can be improved as a whole; The invention also discloses an IO processing device, a terminal and a computer-readable storage medium in a storage system, which have the beneficial effects mentioned above.

Description

technical field [0001] The present invention relates to the technical field of servers, in particular to an IO processing method, device, terminal and computer-readable storage medium in a storage system. Background technique [0002] With the advent of the era of big data, the requirements for high load and low latency of storage devices are getting higher and higher, which requires storage devices to have sufficiently high IOPS (Input / Output Operations Per Second, read and write per second ( The number of I / O) operations), thus supporting the growing massive data business. At present, for some specific scenarios (such as IO overlapping scenarios), for IOs in the same GRAIN, in order to prevent the problem of inconsistency between reading and writing, IO processing is generally performed by queuing, but the queuing method seriously reduces the storage capacity of the storage device. performance, and cannot take advantage of concurrency, resulting in a reduction in the over...

AI Technical Summary

Problems solved by technology

At present, for some specific scenarios (such as IO overlapping scenarios), for IOs in the same GRAIN, in order to prevent the problem of inconsistency between reading and writing, IO processing is generally performed by queuing, but the queuing method seriously reduces the storage capacity of the storage device. performance, unable to take advantage of concurrency, which reduces the overall IOPS of the storage device

Images