An application-aware intelligent cache replacement method and system

Abstract

This invention relates to the field of data processing technology, specifically disclosing an application-aware intelligent cache replacement method and system. The method includes: sensing the cache status of an application scenario and determining candidate cache replacement strategies based on cache replacement strategy matching rules; collecting data access patterns, cache access status, capacity status, and power consumption status data of storage nodes in the application scenario, and performing data preprocessing to generate historical data; training a GRU model based on the historical data to predict future cache hit information and determine the cache hit rate of the application scenario; calculating the strategy effectiveness based on the cache hit rate and its evaluation metrics, and comparing the candidate cache replacement strategies with the current cache replacement strategy; when the cache hit rate is lower than a preset threshold and the strategy effectiveness comparison result exceeds a set value, configuring the candidate cache replacement strategy as the current cache replacement strategy.

Images