Research on high concurrency key technologies for public cloud storage and its system implementation

Research on high concurrency key technologies for public cloud storage and its system implementation

South of the bridge

First of all, this article describes two key technologies that can provide concurrent capabilities: Load balancing and reverse proxy caching. This paper takes LVs as an example to study the algorithms and working modes commonly used in load balancing, and briefly studies the implementation principle of reverse proxy caching. Secondly, this paper analyzes and studies the object storage Swift project of OpenStack, establishes a multi node public cloud storage model based on load balancing, and improves the concurrency performance linearly by extending the number of proxy nodes horizontally. At the same time, considering the high frequency access hotspot storage objects in the system, The paper also proposes a public cloud storage model based on hotspot caching, which forwards the request to the reverse proxy cache, thus reducing the internal pressure of the storage and improving the response speed of the system. Then, a high concurrent public cloud storage System framework based on the load-balanced hotspot cache Public cloud storage model is proposed, and the composition of each functional module and its concrete implementation method are designed in detail. Among them, the function of the original proxy server to judge the hot storage object is improved, so that the public cloud storage system can automatically recognize the request to get the Hotspot storage object and forward it to the cache processing. The test results show that the concurrent processing ability of the high concurrent public cloud storage system is increasing linearly with the increase of the simulated concurrency in the multiuser access scene, while in the simulation Hotspot storage object access scene, its TPS (throughout per Second) than the single agent node of the public cloud storage system increased by 8.48%, than the multi-agent node of the public cloud storage system increased by 4.29%. In summary, the prototype system based on the proposed model can effectively improve the concurrency performance of the public cloud storage service.

Research on high concurrency key technologies for public cloud storage and its system implementation