Data reading method and device for appending, equipment and medium

Abstract

The invention discloses a data reading method and device for appending, equipment, a medium and a program product, and relates to the field of cloud computing, in particular to a cloud storage technology. The method in a specific implementation scheme comprises: in response to the index information corresponding to a logic address in a data reading request in a memory belonging to a secondary index and the secondary index hitting a target cache, acquiring a primary index corresponding to the logic address from the secondary index stored in the target cache; reading data corresponding to the logic address from a disk according to the primary index, wherein the target cache is used for storing the secondary indexes of the appended data and at least two primary indexes corresponding to each secondary index. According to the method, the number of times of twice disk reading when the data reading request hits the secondary index can be reduced, the reading delay of a user is shortened, and the data reading speed of the appending storage engine is increased.

Description

technical field [0001] The present application relates to the field of cloud computing, in particular to a cloud storage technology, and in particular to a data reading method, device, device, medium and program product for additional writing. Background technique [0002] At present, the append-write performance of mainstream non-volatile storage media (SSD / HDD) is much higher than that of random write. Therefore, the append-write storage system can often give full play to the hardware performance. Since additional writing changes the original data location, it is necessary to maintain a mapping relationship from logical address to physical address (LBA[logical block address]—PBA[Physical block address]), that is, index information. [0003] In the append-write storage engine, the method of sharing data streams with data and index information is adopted, and both data and index information are appended to the storage system to eliminate random disk writes. Specifically, se...

AI Technical Summary

Problems solved by technology

[0004] However, when the read data hits the secondary index, it is necessary to read the index information pointed to by the secondary index from the storage system first, and then read the user data from the storage system according to the read index information, and subsequent read requests When the secondary index is hit again, the process needs to be repeated

Therefore, a user read request requires two random disk reads, which greatly increases the delay of the read request.

Images