GC flow control method and device, computer readable storage medium and electronic equipment

Abstract

The invention discloses a GC flow control method and device, a computer readable storage medium and electronic equipment. The method comprises the steps of: presetting a main token bucket, a reservedtoken bucket, token water levels and first storage spaces corresponding to tokens; when the recovery space when GC is completed is received, determining the token number corresponding to the recoveryspace according to the first storage space; adding the tokens of the token number into the main token bucket or the reserved token bucket according to the token water level; and receiving a write IO request, and obtaining a corresponding number of tokens from the main token bucket according to the data size corresponding to the write IO request. According to the embodiment of the invention, the GCrecovery space is stored and used by using the token bucket mechanism, it is ensured that the GC space recovery speed is matched with the speed of the host IO consumption space, and meanwhile, the reserved token bucket and the auxiliary main token bucket are additionally arranged to store tokens generated by the GC recovery space, so that performance jitter generated by different super blocks ofGC can be balanced, and therefore, the stability of IO writing performance of the host is improved.

Description

technical field [0001] The invention relates to the field of hard disk storage, in particular to a GC flow control method, device, computer-readable storage medium and electronic equipment. Background technique [0002] During the use of SSD (Solid State Drives, solid state drives), when the amount of writing reaches the capacity that the host can recognize after the disk is mapped to the operating system and then continues to write, the SSD needs to perform GC (GarbageCollection, garbage collection). )deal with. Such as figure 1 , 2 As shown, the internal storage space of the SSD is composed of multiple super blocks, wherein one data block block is composed of multiple data pages. [0003] With the writing of the host IO, for the same LBA (Logic Block Address, logical data block address) position, if an overwrite occurs, the firmware (firmware) will allocate new space for the overwritten data, and The data stored in the previous space becomes old data, and the space bec...

AI Technical Summary

Problems solved by technology

If the suspension time is too long, it may cause the host to fail to write IO timeout, resulting in host business interruption;

[0006] For the second method, although the space recovery of GC and the space consumption of host IO can be controlled, there will be the following problems: First, the delay control is not accurate enough (the host writes IO in the process of sending it to NAND. After some logical processing, the delay of this processing is uncertain); secondly, because the speed of reclaiming each super block is different (the amount of data that needs to be relocated in each super block is different), therefore, Due to the inability to accurately match GC and host IO, the performance of host IO will fluctuate under this flow control when writing host IO, which is unfavorable for some host applications that require performance stability

Images