A cache optimization method and system based on access frequency

Abstract

The invention discloses a cache optimization method and system based on access frequency, and relates to the technical field of server operation and maintenance management. The method includes: reducing the space required to maintain the cache frequency by adopting a frequency approximate statistical method; and then updating the access frequency of the cache based on the time change. The method reduces the space required for maintaining the cache frequency by adopting the frequency approximate statistical method, and solves the problem of consuming a large amount of computing and storage resources in maintaining high-frequency access metadata, and then uses a statistical method that is updated with the access frequency, The access frequency of the cache is updated based on the time change, which reduces the space occupation of frequency storage, and solves the problems of space consumption under high load and inaccuracy of some data existing in the existing LRU cache design. Based on the access frequency elimination strategy, the invention adopts a cache mechanism with low occupation and high hit rate, thereby reducing resource consumption and improving the cache hit rate.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of server operation and maintenance management, and in particular, to a cache optimization method and system based on access frequency. Background technique [0002] Caching is one of the most basic and effective ways to improve system performance. It achieves fast access by keeping a small number of data items in memory. The intuitive reason that caching works is that data access in many fields of computer science exhibits a considerable degree of locality. A more formal way to capture this locality is to describe the frequency of access to all possible data items through a probability distribution, And note that in real systems the probability distribution is highly skewed, which means that a few objects are visited more frequently than others. When an access arrives, if the requested data is in the cache, it is called a cache hit, and the ratio of the number of hits to the total nu...

AI Technical Summary

Problems solved by technology

There are two problems in the implementation of this kind of cache: one is the need to maintain the number of data accesses, this part of the data will also occupy memory space, and the implementation is cumbersome; the other is that some scenarios are not applicable, such as cached TV dramas, click on The rate is very high, but it decreases significantly after time passes, such high-frequency data should not be left in the cache

However, under high load, LRU still needs a lot of space to achieve the same cache hit rate as LFU

Images