Cache data sharing method and equipment

Abstract

The objective of the invention is to provide a cache data sharing method and equipment. Concretely, a data access request transmitted by the agented process and related to a data file is acquired at a cache agent node side, and the data file is transmitted to the agented process according to the data access request; the process on one or multiple computational nodes is managed through the cache agent node, and the data file required to be used by the agented process is saved in the cache of the cache agent node so that the process on the computational nodes does not need to maintain its independent cache space, the specific data file in the cache of the cache agent node can be shared for multiple processes and cache and computational resources can be saved; meanwhile, the agented process does not need to directly establish connection with a server, and only one connection is established with the server for multiple processes under management of the same cache agent node so that large volume of connection to the server caused by the subscription behavior can be reduced and the load of the server can be reduced.

Description

Technical field [0001] This application relates to the computer field, and in particular to a method and device for sharing cached data. Background technique [0002] In large-scale distributed computing systems, in order to speed up access to server-side data, processes in computing nodes usually introduce caches to manage frequently accessed data. In a distributed scenario, server-side data may change at any time. Therefore, in order to ensure the timeliness of the cache maintained by each process, the cache managed by the process needs to be updated from time to time. However, whether each computing node regularly pulls (Pull) data from the server, or the server occasionally pushes (Push) data to the cache of each computing node process, it will bring access pressure to the distributed computing system. [0003] In the distributed coordination system, the process on the computing node can use "cache + subscription" to obtain the latest content of the subscribed data file, and t...

AI Technical Summary

Problems solved by technology

[0004] Therefore, the data subscription behavior will also increase with the increase of the number of processes, which will lead to the following problems: a) Since the subscription behavior of each process is attached to the independent connection to the server established by the process, with the increase of the number of connections increase, it will increase the load on the server, which poses a serious challenge to the performance of the server

b) Due to the increase in server load, the delay in obtaining feedback for a single subscription behavior increases, which in turn affects the timeliness of the process to obtain the latest content of subscription data

c) Since each process has an independent cache, an increase in the number of subscriptions on a single computing node means an increase in the consumption of cache resources and an increase in operations for processing subscription feedback, which in turn consumes the computing resources of the computing node

Images