A hierarchical scheduling method and system suitable for multi-level storage systems

Abstract

The invention discloses a hierarchical scheduling method and system suitable for a multi-level storage system. The method includes: adding a current storage module read from a multi-level storage system according to a key value to be queried into an up-regulation module queue, and updating the current storage module The reading frequency of the module; if the updated reading frequency of the current storage module is greater than the preset multiple of the minimum reading frequency, the current storage module is determined as the current up-regulation module; the existing in the level above the current level is the same as that in the current up-regulation module When the stored data corresponding to the storage module with the same query key value is stored, the storage data with the same query key value in the current upward adjustment module is stored in the corresponding storage module. The invention determines the storage module whose reading frequency reaches the preset value as the up-regulation module, and up-regulates the data in the up-regulation module, improves the read efficiency, reduces the read amplification, ensures that the hotspot data is the hotspot data in the global state, and guarantees The process of optimizing the read path does not cause write amplification.

Description

technical field [0001] The invention relates to the technical field of computer storage, in particular to a hierarchical scheduling method and system applicable to a multi-level storage system. Background technique [0002] With the development and application of new technologies such as cloud services, big data, and artificial intelligence, the requirements for storage systems are getting higher and higher. The existing multi-level storage systems read data by searching for target values ​​according to levels. Return after finding the record. Compared with the general storage system, the multi-level storage system has an advantage of 1 to 2 orders of magnitude in write throughput, but at the same time, its read performance will be greatly reduced. Generally, the read path Optimization is to reduce the non-hot data in the current level to the low-level storage system to achieve the purpose of optimizing the read path. However, because hot data usually only belongs to hot dat...

AI Technical Summary

Problems solved by technology

[0003] In view of this, the embodiment of the present invention provides a hierarchical scheduling method and system suitable for multi-level storage systems, which solves the problem of write amplification while optimizing the read path in the prior art

Images