A method and terminal for reading and writing cache

Abstract

The invention discloses a method and a terminal for caching reading and writing. By setting a read-write cache between a read-only cache and a common cache, in a reading scenario, the read-write cache can be queried when the corresponding data is not queried in the read-only cache. cache, querying the common cache when no corresponding data is found in the read-write cache, and updating the read-only cache and the read-write cache respectively according to the data queried in the common cache; Clear the read-write cache after the normal cache succeeds. Using the read-write cache as middleware, after the ordinary cache is successfully updated, only the read-write cache is cleared first, which can effectively relieve the read-write pressure under high concurrency when the read-write operations occur at the same time, which is different from the In the existing technology of locking data for reading and writing, which leads to waiting for reading and writing, the present invention can realize the separation of reading and writing of cached data, and supports reading and writing of cached data in high concurrency scenarios.

Description

technical field [0001] The invention relates to the technical field of caching, in particular to a method for caching reading and writing and a terminal. Background technique [0002] Usually, cache servers such as redis or memcache are used as the cache layer in the system to provide cache services in the form of clusters or single nodes. In most cases, the system also implements the function of data storage through local cache. When using local memory as a cache, fast reading and writing of data under high concurrency conditions can be achieved. But there is a new problem at this time, that is, when the data needs to be modified under high concurrency, it will cause frequent read and write conflicts in the memory data. [0003] At present, in order to avoid data inaccuracy caused by read-write conflicts, lock processing is usually performed when writing data, but this method will affect the read operation of data. Since it is executed synchronously and serially, when wri...

AI Technical Summary

Problems solved by technology

But there is a new problem at this time, that is, when the data needs to be modified under high concurrency, it will cause frequent read and write conflicts in the memory data

[0003] At present, in order to avoid data inaccuracy caused by read-write conflicts, lock processing is usually performed when writing data, but this method will affect the read operation of data. Since it is executed synchronously and serially, when writing After the data is locked, the process of reading the data will also wait. The data can only be read after waiting for the completion of the data writing. In the case of high concurrency, it will directly lead to thread freezes, program crashes or Serious consequences such as server downtime

Images