System and method for server overload protection

Abstract

The invention discloses an overload protection system and method for a server. The method comprises the steps of (a) receiving a request for a service from a request end; (b) judging whether the current number of server occupation overtime times of the service is larger than a first threshold value, if not, executing the step (c), and if yes, executing the step (f); (c) allowing the request to be sent to the server; (d) acquiring the server occupation time of the request; (e) when the server occupation time of the request is larger than a second threshold value, updating the current number of server occupation overtime times of the service; (f) judging whether the current number of refusing times of the service is larger than a third threshold value, if yes, allowing the request to be sent to the server, and if not, refusing the request to be sent to the server and updating the current number of refusing times.

Description

technical field [0001] The invention relates to the technical field of server management, in particular to a server overload protection system and method. Background technique [0002] After years of development, there have been many studies on server overload protection, but in actual business scenarios, the server load may still reach its upper limit in some extreme cases. [0003] Among the existing service protection technologies, some start from the hardware, optimize the network card driver, and measure the load of the server according to the parameters such as the CPU usage rate and memory usage rate of the current machine; if the server exceeds the load limit, the network card driver rejects all establishments. A request for a new TCP connection to relieve the pressure on the server. Although optimizing the network card driver can cache the server load to a certain extent, it will cause the throughput of the server to fluctuate for sudden short-term high traffic con...

AI Technical Summary

Problems solved by technology

However, the service protection technology based on the time sliding window ignores the periodicity of visits. In fact, the visits of Internet users will show periodic peaks and valleys. The time window of the

Moreover, the sliding windows of different lengths are only distinguished according to the resources consumed by the services. The granularity is too coarse. If there are multiple services that are CPU-consuming services, it is not convenient to detect the server load in real time. There are load jitters caused by emergencies. Potential for server processing core services to be impacted

Images