Solid-state flash memory write cache system and method based on dynamic non-covering RAID technology

Abstract

The invention discloses a solid-state flash memory write cache system and method based on the dynamic non-covering RAID technology. According to the system, multiple solid-state flash memory hard disks form an RAID to serve as memory space of write cache, and the solid-state flash memory hard disks are connected with a write cache controller; the controller is connected with a front-end IO and back-end storage data, comprises a CPU and a nonvolatile memory module and is used for receiving front-end read requests, reading data from the solid-state flash memory hard disks or back-end storage equipment and receiving front-end write requests, the data are assembled into RAID strips in the nonvolatile memory module and then written in newly distributed strips in the solid-state flash memory hard disks in a non-covering mode, active data in solid-state flash memory strips are written in the back-end storage equipment, and the recovery of the solid-state flash memory strips is achieved. According to the solid-state flash memory write cache system and method based on the dynamic non-covering RAID technology, the strips are assembled dynamically, the phenomenon that the strips are updated frequently is avoided, the system write-in bandwidth is effectively improved, the wear of Flash chips is reduced, and the advantages of being large in capacity and high in property and reliability and the like are achieved.

Description

technical field [0001] The invention belongs to the technical field of data processing, and specifically relates to a large-capacity, high-performance, high-reliability write cache system and method using a solid-state flash memory hard disk, which can be used not only for traditional disk arrays to replace battery-protected DRAM write caches, but also for large-capacity caches cloud storage system. Background technique [0002] Traditional disk arrays rely on RAID, that is, Redundant Array of Inexpensive Disks. The Chinese abbreviation is Redundant Array of Inexpensive Disks. RAID is a combination of multiple independent rotating mechanical hard disks in different ways to form a hard disk group (logical hard disk), thereby providing higher storage performance and data backup technology than a single hard disk. The different ways to form a disk array are called RAID levels (RAID Levels). The disk system utilizes RAID technology to perform parallel access of multiple disks...

AI Technical Summary

Problems solved by technology

However, RAID 6 needs to allocate more disk space for parity information, and has a greater "write loss" compared to RAID 5, so the "write performance" is very poor

Poor performance and complex implementations make RAID 6 rarely used in practice

[0006] It can be seen from the above that one of the main disadvantages of RAID technology is that the write performance of the system is not high enough, especially the random write performance

First, although the rotating mechanical hard disk has a large capacity, its single IO delay is large, at the level of 10mm, much higher than the delay of CPU and DRAM memory

Second, the IOPS performance of the rotating mechanical hard disk is limited, usually a single disk only supports 100 to 200 random read and write operations

[0007] Using DRAM as a write cache for disk arrays has several disadvantages: First, capacity is limited by power consumption and battery protection

There are three disadvantages of this approach: First, the IOPS performance of the hardware RAID card cannot fully match multiple solid-state flash drives, which becomes a new performance bottleneck

Second, hardware RAID will bring additional data read and write due to the maintenance of Parity. For example, for RAID5, writing a small data block will generate 2 read and 2 write operations

Third, hardware RAID needs to use a battery protection module, which brings hidden dangers to reliability

Images