A Real-time Cloud Service Bottleneck Detection Method Based on Kernel Density Estimation and Fuzzy Reasoning System

Abstract

The present invention proposes a real-time cloud service bottleneck detection method based on kernel density estimation and fuzzy reasoning system, comprising the following steps: step 1, start the real-time cloud environment, and run cloud service; step 2, start the real-time cloud service state parameter acquisition component; Step 3. Obtain the cloud service state parameters under the normal operating environment of the cluster for a certain period of time, model it with the kernel density estimation model, and start the fuzzy reasoning system at the same time; Step 4. Input the newly obtained state parameters into the fuzzy reasoning system , calculate the bottleneck index; step 5, set the threshold, observe the historical law of each cloud service bottleneck index, if the threshold has been exceeded within a certain period of time, start the early warning mechanism. The method of the present invention can obtain the running state of each service in the cloud cluster in real time and store it in the data warehouse; realize the bottleneck detection of each real-time cloud service; when the bottleneck index of the real-time cloud service exceeds a certain threshold for a long time, start the early warning mechanism .

Description

technical field [0001] The invention relates to the fields of cloud computing big data computing, real-time service computing and anomaly detection, in particular to a real-time cloud service bottleneck detection method based on a kernel density estimation and fuzzy reasoning system. Background technique [0002] Anomaly detection of real-time services in cloud computing environment is extremely difficult. [0003] The first is the difficulty in obtaining and selecting cloud service status parameters. Cloud services in a real-time cloud environment are distributed in each cluster node, and some cloud services are still running on different virtual machines on the same node. Status is hard to get. [0004] Secondly, the operation mode of real-time cloud services is unpredictable, and there is no standard to define the idle and busy states of a cloud service, which makes it difficult to define the bottleneck of a cloud service. In addition, the scale of the cluster is uncont...

AI Technical Summary

Problems solved by technology

[0002] Anomaly detection of real-time services in cloud computing environment is extremely difficult

[0003] The first is the difficulty in obtaining and selecting cloud service status parameters. Cloud services in a real-time cloud environment are distributed in each cluster node, and some cloud services are still running on different virtual machines on the same node. status hard to get

[0004] Secondly, the operation mode of real-time cloud services is unpredictable, and there is no standard to define the idle and busy states of a cloud service, which makes it difficult to define the bottleneck of a cloud service

In addition, the scale of the cluster is uncontrollable. When the number of nodes in a cloud computing cluster is large, this creates obstacles to the network within the cluster, and obtaining the status of the cloud service itself consumes resources, which also causes problems for cloud services. The bottleneck detection algorithm must require low resource consumption and other characteristics; at the same time, a large number of cloud services also require that the detection algorithm must be unsupervised, otherwise it will consume a lot of manual operations, and the latter is obviously not feasible

Images