Research on Load Balancing Strategy of Standalone Multi-core Cluster based on Node.js
Bo Wu, Zhengping Jin · 2015
With the development of Internet and the rapid increase of users, there are a lot of serious problems in Internet, such as network congestion, server overload and too long response time.So it is necessary for network applications to support the high concurrent access requirements.Node.js is an asynchronous and non-blocking I/O model that provides efficient scheme for this problem.However, as for the Node.jsnative single-threaded work pattern cannot use performance of multicore CPUs, the advantage of high concurrency cannot given full paly to.In this paper, we proposed a simple and efficient Node.js multi-core and parallel framework based on the study of the existing multi-core solutions.This framework balances the nodes through the way of request classification and weighted round-robin.The strategy consider the difference between the user requests and the working process of real-time load totally.Moreover, at the same time we introduce the load feedback mechanism, which can balance the load of nodes and make full use of the resources of each node better through allocating user request according to equivalent transformation of the load weight and accelerated decline weights of critical state.Experiment show that the plan has better load balance ability compared to the the original solutions under the condition of high concurrent requests.