Write handling method for disc array arrangement

Abstract

This invention relates to a written process method in a disk array structure, which writes data into the cache if it is not full when the written request of the host arrives and pre-writes back to full-fill the branches where the written request in the cache places and several branches after that. When the cache is full, the host written requirement waits and selects some from the pre-written zones to write back and takes one from the waiting queue of the host when a region finishes the written back to try the host written, which reduces vacancies and number of written requirement to increase the merge possibility to increase the merge possibility and the written-back efficiency of cache in the storage system.

Description

technical field [0001] The invention relates to data storage technology, in particular to a write processing method in a disk array structure. Background technique [0002] With the continuous expansion of computer application fields, users have higher and higher requirements for storage system capacity, reliability and availability, and faster speed. The processing speed of the microprocessor is improved rapidly, but the speed of the disk as the main external storage is not improved much due to the limitation of mechanical movement (the access time stays at the millisecond level), which greatly limits the overall performance of the storage system. The disk array structure (RAID) with Cache (caching) to solve this bottleneck occupies a dominant position in high-performance computer storage due to its advantages of high efficiency, high capacity, high reliability, flexibility and low price. [0003] Disk array technology forms ordinary hard disks into a disk array. When the ...

AI Technical Summary

Problems solved by technology

[0011] Although the above-mentioned technical scheme improves the write-back efficiency to a certain extent, because it is "the data from the target data block to the end of the track where it is read into the Old Data", if the new data written is less than one minute According to this prefetching method, the stripe where the new data is located will form a "hole", that is, there is no "dirty data" in the Cache (dirty data refers to the data in the Cache that is inconsistent with the disk, that is, the data that the host writes to the Cache data, but the data has not been synchronized to the disk), and there is no "old data" (old data refers to the data in the Cache that is consistent with the disk, that is, the data that has been written by the host to the disk, or read from the disk to the Cache data), that is to say, a data block in this stripe has never been written by the host, or has never been covered by disk data, that is, there is a "hole"

[0015] Since the total capacity of the Cache is limited, according to this write-back method, since each write request needs to read the data from the target data block to the end of the track where it is located, although the write-back method is simplified, it is difficult to make better use of it. Cache limited space

In this way, when the Cache writes back, the data in the "hole" stripe will have to be split into multiple write requests, which increases the number of write requests and reduces the write-back efficiency

Images