A method for end-to-end fault monitoring of services in a cloud computing environment

Abstract

The present invention relates to a method for end-to-end fault monitoring of services in a cloud computing environment. Based on real-time monitoring of four dimensions, namely, cloud pool, network, equipment, and services, and combined with static and dynamic threshold algorithms, relevant faults in each dimension are monitored in real time, and a fault monitoring system for three layers, namely, IaaS, PaaS, and SaaS, is established to analyze and diagnose faults in an all-round way, thereby providing an accurate means for quickly locating service faults, improving the service quality of the service, and improving customer satisfaction.

Description

Technical Field

[0001] The present invention relates to the field of deploying probes and integrating acquisition programs in a cloud computing environment, and in particular to a method for end-to-end fault monitoring of services in a cloud computing environment. Background Art

[0002] With the rapid development of cloud computing, many telecom operators have moved many of their services to the cloud. The cloud computing environment gathers a large number of physical and virtual resources and provides services at multiple levels, such as IaaS, PaaS and SaaS. Failures at the IaaS and PaaS layers often lead to failures at the SaaS layer, and there may be multiple links or nodes at each layer of IaaS, PaaS and SaaS. Due to the diversity of cloud computing service data and the dynamic nature of the deployment environment, when the host fails or there is a performance bottleneck, cloud computing nodes often migrate dynamically from one host to another, which can easily lead to abnormalities, and abnormalities usually turn into serious failures, causing service failures. Moreover, abnormal nodes during operation may cause abnormalities in other nodes associated with them, and further cause large-scale service failures. On the other hand, the cloud host of the collection node may also misjudge the failure due to its own dynamic nature, resulting in failure of external network connection, causing interference and affecting the credibility of the alarm. At present, when a service failure is found, it is often necessary to first check from the business service end to the end physical resource node, which requires cross-departmental collaboration, slow fault location, and long fault handling. Therefore, abnormal monitoring and correct fault judgment of related nodes in cloud computing environment is an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide a method for end-to-end fault monitoring of services in a cloud computing environment, so as to enhance the efficiency of locating and processing service faults in the cloud computing environment and improve the accuracy of fault diagnosis.

[0004] To achieve the above object, the technical solution of the present invention is: a method for end-to-end fault monitoring of services in a cloud computing environment, comprising the following steps:

[0005] Step S1: Deploy the integrated acquisition service to connect to the cloud resource pool through the API interface to obtain the cloud host, host machine and computing resource pool information that carries the business on the cloud resource pool;

[0006] Step S2: Deploy collection probes on the host machines in the same computing resource pool to collect IaaS layer performance monitoring indicators on each host machine, and send them to the integrated collection service. The integrated collection service determines whether there is an alarm based on the defined threshold;

[0007] Step S3: deploy a collection probe on each cloud host to collect IaaS layer performance indicators of each cloud host;

[0008] Step S4: Deploy a collection probe on the cloud host where the PaaS component is installed, collect the performance indicators of the PaaS component, and send them to the integrated collection service. The integrated collection service determines whether there is an alarm based on the defined threshold;

[0009] Step S5: Establish a cross-collection matrix to monitor the network quality of the host and cloud host in real time online, and make alarm determinations and early warnings based on the principles of time correlation and alarm consistency;

[0010] Step S6: deploy an integrated collection service to collect business service status indicators SaaS, and determine whether there is an alarm based on the threshold;

[0011] Step S7: deploy a collection probe on each cloud host that carries the service, collect the service status indicator SaaS of the other end between adjacent nodes, and then send the data to the integrated collection service, which determines whether an alarm is required based on the threshold;

[0012] Step S8: Draw the end-to-end full-process display nodes of the service from bottom to top, and project the alarms and performance indicators or business indicators of each node.

[0013] Compared with the prior art, the present invention has the following beneficial effects: the present invention can perform comprehensive fault analysis and diagnosis, thereby providing an accurate means for quickly locating business faults, improving the service quality of the business, and improving customer service satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of end-to-end fault monitoring, location collection, and monitoring and early warning in the cloud computing environment of the present invention.

[0015] Figure 2 This is a flow chart of a cross-monitoring matrix for end-to-end fault monitoring and location in a cloud computing environment of the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0017] like Figure 1 , 2 As shown, the present invention provides a method for end-to-end fault monitoring of services in a cloud computing environment, comprising:

[0018] 1. If Figure 1 As shown in the figure, the end-to-end fault monitoring, location collection, monitoring and early warning of services in a cloud computing environment are implemented in the following specific logic:

[0019] (1) Deploy probes on the host machines of the computing resource pool collected by the resource pool API interface to collect the performance indicators of the host machines, including CPU utilization, memory utilization, file system space utilization, and time, to monitor whether the host machines can run normally.

[0020] (2) Install IaaS layer probes on the cloud host that carries the business and collect performance indicators of the cloud host, including CPU utilization, memory utilization, file system space, and collection time, to monitor whether the cloud host that carries the business can provide business services normally.

[0021] (3) Install probes on the cloud host that hosts the PaaS service to collect performance indicators of the PaaS components, such as the table space utilization of the database service and the request waiting time of the message queue.

[0022] (4) Collect the status of business services through integrated collection services, such as service response status, service response duration and time.

[0023] (5) Install probes on each cloud host node that carries the SaaS business service to collect the status and time of the service at the other end of the node.

[0024] (6) The integrated collection service automatically schedules each collection task, receives the indicator data collected by the probe, and then calls the threshold engine to monitor whether the indicator data exceeds the threshold. If the threshold is exceeded, alarm data is generated.

[0025] (7) Store, filter and process the generated alarm data.

[0026] (8) Based on the logical end-to-end relationship of the business, an end-to-end visual topology is established, and the alarms of the corresponding nodes are projected onto each topological node to present them in a more intuitive way.

[0027] 2. If Figure 2 As shown, the specific implementation steps of the cross-collection matrix of service network quality in the cloud computing environment are as follows:

[0028] (1) Deploy an integrated collection service on each of the three cloud hosts in different network segments to establish a cross-collection matrix.

[0029] (2) The three integrated collection services initiate network quality monitoring of the target host and cloud host through ICMP.

[0030] (3) The three integrated collection services dispatch the threshold engine based on the monitoring strategy to determine whether the network quality exceeds the threshold and issue an alarm.

[0031] (4) Based on the time correlation, the network quality alarms collected by the three integrated collection services are judged to be related. According to the alarm consistency principle, when the three collection services all issue network quality alarms at similar times, the alarm is judged to be a real alarm; otherwise, it is judged to be a fault of the integrated collection service that currently generates the alarm.

[0032] (5) Store, filter and process the network quality alarm data that is judged to be

[0033] (6) Project the alarm onto the end-to-end visualization topology device of the service, and present it in a more eye-catching way with the color identification that is different from the performance indicator alarm, so as to quickly determine the location and type of the fault. Table 1 is a table related to fault classification, fault type and indicators.

[0034] Table 1

[0035] Fault classification Fault type index Performance indicator alarm IaaS CPU utilization, memory utilization, file system space utilization, etc. Network quality indicator alarm IaaS Network connectivity, network packet loss rate, network latency, etc. Database performance indicators PaaS Table space utilization, invalid index, database monitoring status, session occupancy, etc. Message Queue Performance Indicators PaaS Waiting time in the request queue, node processing time, response time, etc. Business end-to-end service performance SaaS Service port connectivity, etc. Business Indicators SaaS Call connection rate, etc.

[0036] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.

Claims

1. A method for end-to-end fault monitoring of services in a cloud computing environment, characterized in that: The steps include: Step S1: Deploy the integrated acquisition service to connect to the cloud resource pool through the API interface to obtain the cloud host, host machine and computing resource pool information that carries the business on the cloud resource pool; Step S2: Deploy collection probes on the host machines in the same computing resource pool to collect IaaS layer performance monitoring indicators on each host machine, and send them to the integrated collection service. The integrated collection service determines whether there is an alarm based on the defined threshold; Step S3: deploy a collection probe on each cloud host to collect IaaS layer performance indicators of each cloud host; Step S4: Deploy a collection probe on the cloud host where the PaaS component is installed, collect the performance indicators of the PaaS component, and send them to the integrated collection service. The integrated collection service determines whether there is an alarm based on the defined threshold; Step S5: Establish a cross-collection matrix to monitor the network quality of the host and cloud host in real time online, and make alarm determinations and early warnings based on the principles of time correlation and alarm consistency; Step S6: deploy an integrated collection service to collect business service status indicators SaaS, and determine whether there is an alarm based on the threshold; Step S7: deploy a collection probe on each cloud host that carries the service, collect the service status indicator SaaS of the other end between adjacent nodes, and then send the data to the integrated collection service, which determines whether an alarm is required based on the threshold; Step S8: Draw the end-to-end full-process display nodes of the service from bottom to top, and project the alarms and performance indicators or business indicators of each node; In step S5, the cross acquisition matrix is ​​specifically implemented as follows: (1) Deploy an integrated collection service on each of the three cloud hosts in different network segments to establish a cross-collection matrix; (2) The three integrated collection services initiate network quality monitoring of the target host and cloud host through ICMP respectively; (3) The three integrated collection services dispatch the threshold engine based on the monitoring strategy to determine whether the network quality exceeds the threshold and issue an alarm; (4) Based on the time correlation, the network quality alarms collected by the three integrated collection services are judged to be related; based on the alarm consistency principle, when the three integrated collection services all issue network quality alarms at similar times, the corresponding alarms are judged to be real alarms; otherwise, it is judged to be a fault of the integrated collection service that currently generates the alarm; (5) Store and filter the network quality alarm data that is judged to be; (6) Project the alarm onto the end-to-end visualization topology device of the service to quickly determine the location and type of the fault.

Images