CDN-accelerated domain name lag analysis methods and computer program products
The CDN-accelerated domain name lag analysis method, which uses multi-dimensional indicator analysis and dynamic periodic adjustment, solves the problem of not being able to quickly locate domain name anomalies in existing technologies, and achieves efficient and accurate lag problem investigation and handling.
Patent Information
- Application Number
- CN202611610603.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-19
- Publication Date
- 2026-11-17
AI Technical Summary
Existing CDN-accelerated domain name analysis methods cannot quickly locate specific domain name anomalies, rely on a single indicator and are prone to false positives and false negatives, lack multi-dimensional diagnosis and automated analysis, resulting in low investigation efficiency and highly subjective results.
By combining multi-dimensional metrics, including first-byte time, interruption rate, and business bandwidth, abnormal CDN nodes are screened, and the operating status parameters of nodes, switches, and hosts are obtained. The analysis cycle is dynamically adjusted to achieve accurate positioning from the node level to the domain level.
It enables proactive detection and timely handling of CDN-accelerated domain name lag issues, reduces false positives and false negatives, improves investigation efficiency and accuracy, and reduces system resource consumption.
Smart Images

Figure CN123123347A_ABST
Abstract
Description
Technical Field
[0001] This disclosure generally relates to the field of network communication technology, and in particular to a CDN-accelerated domain name lag analysis method and computer program product. Background Technology
[0002] Content Delivery Network (CDN) accelerated domains frequently experience stuttering during operation, manifesting as video playback buffering, slow loading, or connection interruptions, severely impacting user experience. Causes of this stuttering include, but are not limited to, excessive node load, low cache hit rate, poor network link quality, switch bandwidth congestion, and single-machine hardware failure.
[0003] Currently, CDN service providers typically deploy two independent systems: a monitoring center and a log center. The monitoring center is responsible for collecting node performance metrics of CDN nodes and using static threshold rules to determine whether CDN nodes are abnormal and trigger alarms. The log center is responsible for recording business access logs and statistically analyzing information such as requests per second and traffic distribution. However, the node performance metrics of the monitoring center and the business traffic data of the log center belong to two independent systems, and there is no correlation analysis between the two.
[0004] Existing methods for analyzing domain name lag have the following shortcomings: First, the monitoring center can only identify node-level anomalies and cannot quickly pinpoint the anomaly to the specific accelerated domain. After receiving node alarms, operations and maintenance personnel still need to manually perform secondary analysis in conjunction with business data from the log center, resulting in low troubleshooting efficiency. Second, relying solely on a single indicator for anomaly detection is prone to false alarms or missed alarms, lacking multi-dimensional comprehensive diagnostic capabilities. Third, the analysis of the causes of lag issues depends on the personal experience of operations and maintenance personnel, lacks an automated cause analysis mechanism, and the diagnostic results are highly subjective and have poor reproducibility. Summary of the Invention
[0005] In view of the aforementioned defects or deficiencies in the existing technology, this application aims to provide a CDN-accelerated domain name lag analysis method and computer program product. If the current time is the trigger time of a preset period, the method determines whether an abnormal node exists based on the first byte time, interruption rate, and service bandwidth of each CDN node. If an abnormal node exists, the method analyzes and obtains the root cause of lag in the accelerated domain name of the abnormal node based on the node granularity indicators, switch granularity indicators, host granularity indicators, and traffic distribution granularity indicators of the abnormal node. The preset period is then reset for the first time to determine whether the new current time is the trigger time of the new preset period. If no abnormal node exists, the preset period is reset for the second time to determine whether the new current time is the trigger time of the new preset period. This application achieves rapid location of lag from node abnormality to domain name lag through multi-dimensional indicator joint analysis. By dynamically adjusting the analysis period, it improves the efficiency of lag analysis and the utilization rate of system resources.
[0006] Firstly, this application provides a method for analyzing CDN-accelerated domain name buffering issues. The method includes: If the current time is the trigger time of the preset period, the first byte time, interruption rate and service bandwidth of each CDN node are obtained, and based on the first byte time, interruption rate and service bandwidth, it is determined whether there are abnormal CDN nodes among the CDN nodes. In the case where an abnormal CDN node exists among the CDN nodes, the node operating status parameters, switch operating status parameters, host operating status parameters, and multi-dimensional traffic distribution parameters of the abnormal CDN node are obtained. Based on the node operating status parameters, switch operating status parameters, host operating status parameters, and / or the multi-dimensional traffic distribution parameters, a stuttering analysis is performed on the accelerated domain name of the abnormal CDN node to obtain the stuttering analysis result of the domain name corresponding to the current time. Then, the preset period is reset for the first time, and the period after the first reset is used as the new preset period. It is then determined whether the new current time is the trigger time of the new preset period; the period after the first reset is less than the preset period. If no abnormal CDN node is found among the CDN nodes, the preset period is reset for the second time, and the period after the second reset is used as the new preset period. It is then determined whether the new current time is the trigger time of the new preset period. The period after the second reset is greater than or equal to the preset period.
[0007] In conjunction with the first aspect, in one possible implementation, determining whether there are abnormal CDN nodes among the CDN nodes based on the first byte time, the interruption rate, and the service bandwidth includes: For each CDN node, determine whether the first byte time of the CDN node is higher than a first preset time threshold, whether the interruption rate of the CDN node is higher than a first interruption rate threshold, and whether the service bandwidth of the CDN node is higher than a first preset bandwidth threshold. If the first byte time is higher than the first preset time threshold, or the interruption rate is higher than the first preset interruption rate threshold, or the service bandwidth is higher than the first preset bandwidth threshold, then the CDN node is determined to be the abnormal CDN node. Among them, CDN nodes whose service bandwidth is lower than the second preset bandwidth threshold are not identified as abnormal CDN nodes, and the first preset bandwidth threshold is greater than the second preset bandwidth threshold.
[0008] In conjunction with the first aspect, in one possible implementation, the step of performing a stuttering analysis on the accelerated domain name of the abnormal CDN node based on the node operating status parameters, the switch operating status parameters, the host operating status parameters, and / or the multi-dimensional traffic distribution parameters, to obtain the domain name stuttering analysis result corresponding to the current moment, includes: If the retransmission rate of the abnormal CDN node is higher than the preset retransmission rate threshold, and the switch bandwidth of the abnormal CDN node is higher than the preset switch bandwidth threshold, and the switch health value of the CDN node is lower than the preset switch health threshold, then it is determined that the node running status parameters and the switch running status parameters meet the first judgment rule, and it is determined that the domain name lag analysis result is that the performance problem is caused by the switch corresponding to the abnormal CDN node running at full capacity. If the node operating status parameters and the switch operating status parameters do not meet the first judgment rule, and the node retransmission rate is higher than the preset node retransmission rate threshold, and the switch bandwidth is lower than the preset switch bandwidth threshold, and the switch health value is higher than the preset switch health threshold, then it is determined that the node operating status parameters and the switch operating status parameters meet the second judgment rule, and it is determined that the domain name lag analysis result is a performance problem caused by hardware failure of the host group corresponding to the abnormal CDN node. The node retransmission rate is a node operating status parameter, and the switch bandwidth and the switch health value are switch operating status parameters.
[0009] In conjunction with the first aspect, in one possible implementation, the method further includes: If the node operating status parameters and the switch operating status parameters do not meet the second judgment rule, and the node packet loss rate of the abnormal CDN node is higher than the preset node packet loss rate threshold, and the host average service bandwidth of the abnormal CDN node is higher than the preset host average bandwidth threshold, and the host average health value of the abnormal CDN node is lower than the preset host average health threshold, then it is determined that the node operating status parameters and the host operating status parameters meet the third judgment rule, and it is determined that the domain name lag analysis result is that the abnormal CDN node is overloaded, causing performance problems. The node packet loss rate is a node operating status parameter, and the host average service bandwidth and the host average health value are host operating status parameters.
[0010] In conjunction with the first aspect, in one possible implementation, the method further includes: If the node running status parameters and the host running status parameters do not meet the third judgment rule, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average disk load of the abnormal CDN node is higher than the preset host average disk load threshold, then it is determined that the node running status parameters and the host running status parameters meet the fourth judgment rule, and it is determined that the domain name lag analysis result is that the low hit rate of the abnormal CDN node caused the performance problem. Among them, the node hit rate and the node back-to-parent / back-to-source bandwidth are node operating status parameters, and the host average disk load is host operating status parameter.
[0011] In conjunction with the first aspect, in one possible implementation, the method further includes: If the node running status parameters and the host running status parameters do not meet the fourth judgment rule, and the traffic within a single client-domain node of the abnormal CDN node is higher than the preset traffic threshold or the number of requests per second is higher than the preset number of requests threshold, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average health value of the CDN node is lower than the preset host average health threshold, then it is determined that the multidimensional traffic distribution parameters, the node running status parameters, and the host running status parameters meet the fifth judgment rule, and it is determined that the domain name lag analysis result is a performance problem caused by a sudden increase in the client domain name traffic or the number of requests per second of the abnormal CDN node. Among them, the traffic within a single client-domain node and the number of requests per second belong to the multidimensional traffic distribution parameters, the node back-to-parent / back-to-origin bandwidth and the node hit rate belong to the node operating status parameters, and the host average health value belongs to the host operating status parameters.
[0012] In conjunction with the first aspect, in one possible implementation, the method further includes: If the multidimensional traffic distribution parameters, the node running status parameters, and the host running status parameters do not meet the fifth judgment rule, and the single host first byte time of the abnormal CDN node is higher than the second preset time threshold, the single host interruption rate is higher than the second preset interruption rate threshold, the single host health value is lower than the preset host health threshold, and the single host disk load is higher than the preset host disk load threshold, then it is determined that the host running status parameters meet the sixth judgment rule, and it is determined that the domain name lag analysis result is that the single host disk abnormality of the abnormal CDN node caused the node performance problem; If the host operating status parameters do not meet the sixth judgment rule, and the first byte time of a single host is higher than the second preset time threshold, and the single host interruption rate is higher than the second preset interruption rate threshold, and the single host health value is lower than the preset host health threshold, and the single host CPU load is higher than the preset CPU load threshold, and the single host download speed is lower than the preset download speed threshold, then it is determined that the host operating status parameters meet the seventh judgment rule, and it is determined that the domain name lag analysis result is a node performance problem caused by the single host anomaly of the abnormal CDN node; If the host operating status parameters do not meet the seventh judgment rule, then the domain name lag analysis result is determined to be an unknown performance problem; Among them, the single-host first byte time, the single-host interruption rate, the single-host health value, the single-host disk load, the single-host CPU load, and the single-host download speed are all host operating status parameters.
[0013] In conjunction with the first aspect, in one possible implementation, the method further includes: Based on the node operating status parameters, the switch operating status parameters, the host operating status parameters, the multidimensional traffic distribution parameters, and the domain name lag analysis results, a domain name lag analysis record corresponding to the current moment is generated and stored.
[0014] In conjunction with the first aspect, in one possible implementation, the method further includes: Perform a domain lag analysis record clearing operation at fixed intervals, deleting all domain lag analysis records from before the preset time period.
[0015] Secondly, this application also provides a computer program product. This computer program product includes instructions that, when executed, cause the method described in the first aspect to be implemented.
[0016] This application provides a CDN-accelerated domain name lag analysis method and computer program product. By automatically triggering the lag analysis process at a preset cycle, it can proactively detect performance degradation of CDN node-accelerated domain names. Compared with the existing technology of passively waiting for user complaints before investigation, it realizes proactive detection and timely handling of lag problems. By comprehensively judging three indicators, namely first byte time, interruption rate and business bandwidth, abnormal CDN nodes are initially screened. Compared with the existing technology that relies on only a single indicator for anomaly judgment, it effectively reduces false alarms and false negatives and avoids the computational overhead of full analysis of all CDN nodes. By further acquiring node operation status parameters, switch operation status parameters, host operation status parameters, and multi-dimensional traffic distribution parameters of abnormal CDN nodes to perform stuttering analysis on accelerated domains, the anomaly location can be refined from the node level to the specific accelerated domain. This overcomes the shortcomings of existing technologies that can only identify node anomalies but cannot locate specific domains. Moreover, it eliminates the need for manual cross-system data correlation analysis, improving troubleshooting efficiency. By dynamically adjusting the analysis cycle based on the presence of abnormal CDN nodes, an adaptive balance between domain stuttering analysis efficiency and system performance is achieved, thereby significantly improving the accuracy of CDN accelerated domain stuttering analysis. Attached Figure Description
[0017] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is one of the flowcharts illustrating a CDN-accelerated domain name lag analysis method in one embodiment; Figure 2 This is the second flowchart of a CDN-accelerated domain name lag analysis method in one embodiment; Figure 3 This is the third flowchart of a CDN-accelerated domain name lag analysis method in one embodiment; Figure 4 This is the fourth flowchart of a CDN-accelerated domain name lag analysis method in one embodiment; Figure 5 This is the fifth flowchart of a CDN-accelerated domain name lag analysis method in one embodiment; Figure 6 This is a block diagram of a CDN-accelerated domain name lag analysis system in one embodiment; Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0018] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.
[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of this invention can be combined with each other. The invention will now be described in detail with reference to the accompanying drawings and embodiments. Furthermore, the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The terms "first" and "second," etc., in the specification and claims of the embodiments of this invention are used to distinguish different objects, not to describe a specific order of objects.
[0020] CDN-accelerated domains often experience stuttering during operation, manifesting as video playback buffering, slow loading, or connection interruptions, which seriously affect the user experience. These stuttering issues may be caused by a variety of factors, including but not limited to: (1) excessive node load leading to decreased processing capacity; (2) low cache hit rate leading to increased back-to-origin pressure; (3) poor network link quality causing packet loss and retransmission; (4) switch bandwidth congestion affecting data transmission; and (5) single-machine hardware failure.
[0021] Traditional problem-solving methods mainly rely on maintenance personnel to manually check monitoring charts and analyze log files, which has the following shortcomings: (1) Passive response - usually only after user complaints, lacking a proactive discovery mechanism; (2) Low efficiency - requires manual association of data from multiple monitoring systems, which takes a long time; (3) Difficult diagnosis - lacks automated root cause analysis capabilities, and locating problems requires rich experience; (4) Data silos - performance indicators, business data, and system logs are scattered across different platforms, making unified analysis difficult.
[0022] In existing technologies, the following technical solutions are usually used to analyze the causes of CDN acceleration domain name lag: (1) The monitoring center collects CDN node performance indicators separately, such as first byte time, packet loss rate, retransmission rate, bandwidth utilization rate, traffic hit rate, etc.; (2) The log center collects business access logs separately and counts information such as Queries Per Second (QPS), bandwidth, customers, domain names, operators, and regions; (3) The threshold rules or simple models are used to determine whether a CDN node is abnormal and issue a warning; (4) The operation and maintenance personnel manually check the domain name, data center and host machine based on the alarm.
[0023] The above-mentioned existing technologies have the following shortcomings: (1) They can only judge node-level anomalies and cannot quickly locate the anomalies to the accelerated domain name; (2) They cannot combine the customer / channel / regional traffic distribution of the log center with the node performance indicators of the monitoring center for joint analysis; (3) Relying on a single indicator is prone to false alarms or missed alarms, and lacks multi-dimensional comprehensive diagnosis; (4) They cannot automatically summarize the causes of anomalies and require manual analysis; (5) There is no structured storage and historical query mechanism for the lag analysis records of CDN accelerated domain names.
[0024] To address the aforementioned technical issues, this application provides a CDN-accelerated domain name lag analysis method and computer program product. This method can automatically filter abnormal CDN nodes, comprehensively determine the root cause of lag based on multi-source information such as switches, nodes, and hosts, and generate lag records. It enables filtering queries based on dimensions such as time range, node, customer, domain name, province, and ISP, facilitating operators to quickly locate the cause of domain name lag.
[0025] The following is combined Figures 1 to 5 This application describes a CDN-accelerated domain name congestion analysis method, wherein the execution subject of the CDN-accelerated domain name congestion analysis method is a computer device. This computer device can be a standalone physical server, cloud server, personal computer, or smart terminal, or a combination or cluster of the above devices.
[0026] Reference Figure 1 This is a schematic diagram of the CDN acceleration domain name lag analysis method provided in the embodiments of this application, such as... Figure 1 As shown, the CDN-accelerated domain name lag analysis method includes the following steps 101 to 103.
[0027] Step 101: If the current time is the trigger time of the preset period, obtain the first byte time, interruption rate and service bandwidth of each CDN node, and determine whether there are abnormal CDN nodes in each CDN node based on the first byte time, interruption rate and service bandwidth.
[0028] It should be noted that the method in this embodiment is automatically triggered at a preset period. The initial value of the preset period can be pre-configured, for example, it can be set to 5 minutes. The current time is monitored in real time. If the current time reaches the trigger time of the preset period, such as the 5th minute, 10th minute, 15th minute, etc. after the lag analysis process starts, then a lag analysis process is triggered.
[0029] Upon triggering, the monitoring center first queries the first byte time, interruption rate, and service bandwidth of each CDN node. The first byte time is the time from when a user initiates a request to when the first response byte is received, reflecting the node's response speed; the interruption rate is the proportion of failed requests to total requests per unit time, reflecting the node's service availability; and the service bandwidth is the node's data throughput per unit time, reflecting the node's load level.
[0030] Then, based on the first byte time, interruption rate, and service bandwidth of each CDN node, it is determined whether each CDN node is an abnormal CDN node. In one embodiment, corresponding weights are configured for the three indicators of first byte time, interruption rate, and service bandwidth, and the pre-defined abnormal conditions for each CDN node are calculated. For example, the weight of first byte time is 0.4, the weight of interruption rate is 0.4, and the weight of service bandwidth is 0.2. The normalized score of each indicator is calculated, and then the normalized scores of each indicator are multiplied by their corresponding weights and summed to obtain the comprehensive score of each CDN node. If the comprehensive score of a CDN node exceeds the pre-defined comprehensive score threshold, the CDN node is determined to be an abnormal CDN node; conversely, if the comprehensive score of another CDN node does not exceed the pre-defined comprehensive score threshold, the CDN node is determined to be a normal CDN node.
[0031] In another embodiment, the first byte time, interruption rate, and service bandwidth of each CDN node within a historical time period are collected as sample features. Operations personnel then label each sample feature with either an anomaly or normal flag, forming a training sample set. The first byte time, interruption rate, and service bandwidth of each training sample in the training sample set are then used as input features, and the anomaly or normal flags of these training samples are used as supervision labels. Machine learning algorithms (such as logistic regression, support vector machine, decision tree, random forest, or gradient boosting tree) are then used to train the anomaly detection model, which is then deployed on a computer device.
[0032] When actually executing the stuttering analysis process, the first byte time, interruption rate and service bandwidth of each CDN node obtained at the current moment can be directly input into the anomaly detection model. The anomaly detection model automatically outputs the judgment result of each CDN node as abnormal or normal, and the CDN node judged as abnormal is identified as an abnormal CDN node.
[0033] Step 102: In the case of abnormal CDN nodes among the CDN nodes, obtain the node operation status parameters, switch operation status parameters, host operation status parameters, and multi-dimensional traffic distribution parameters of the abnormal CDN nodes. Based on the node operation status parameters, switch operation status parameters, host operation status parameters, and / or multi-dimensional traffic distribution parameters, perform a stuttering analysis on the accelerated domain name of the abnormal CDN node to obtain the stuttering analysis result of the domain name corresponding to the current time. Then, perform a first reset on the preset period, take the period after the first reset as the new preset period, and determine whether the new current time is the trigger time of the new preset period; the period after the first reset is less than the preset period.
[0034] Specifically, if there are any abnormal CDN nodes among the CDN nodes at the current moment, the monitoring center will query the node operation status parameters of the abnormal CDN nodes, including but not limited to parent / origin bandwidth, hit rate, packet loss rate, retransmission rate, and download speed; simultaneously, the monitoring center will query the operation status parameters of the switch corresponding to the abnormal CDN node, including but not limited to bandwidth and health value; simultaneously, the monitoring center will query the host operation status parameters of each host within the abnormal CDN node, including but not limited to network interface bandwidth, service bandwidth, parent / origin bandwidth, hit rate, packet loss rate, retransmission rate, download speed, first byte time, interruption rate, CPU load, disk load, and health value; and simultaneously, the log center will query the multi-dimensional traffic distribution parameters of the abnormal CDN node, including but not limited to the traffic distribution of the node carried by customers-domain-province-carrier and the number of requests per second within the last M minutes. For example, the value of M is greater than the preset period; when the preset period is 5 minutes, M is 15 minutes; when the preset period is 3 minutes, M is 10 minutes.
[0035] At this point, the node operating status parameters, switch operating status parameters, host operating status parameters, and multi-dimensional traffic distribution parameters of the abnormal CDN node are matched with preset root cause determination rules to determine the specific reason for the lag in the accelerated domain name of the node. In one embodiment, taking node A as an example, its node retransmission rate is 15% (higher than normal), switch bandwidth is 90Gbps (close to full capacity), and switch health value is 60 (lower than normal). According to the preset determination rules, the root cause of the lag in the accelerated domain name of node A is determined to be a performance problem caused by the switch bandwidth being fully utilized, thus obtaining the lag analysis result for the domain name corresponding to node A.
[0036] After the stuttering analysis is completed, the preset period is reset for the first time. Since there are abnormal CDN nodes within the current analysis period, it indicates that the CDN system is experiencing performance degradation, requiring an increase in the analysis frequency to quickly identify and locate more potential problems. Therefore, when abnormal CDN nodes are present, the period is shortened to increase the analysis frequency and facilitate faster response. For example, the preset period is reset from the current 5 minutes to 3 minutes, triggering subsequent stuttering analysis processes at shorter intervals. After the first reset, the current time continues to be monitored to determine if the new current time is the trigger time for the new preset period (i.e., the 3-minute period), so that the stuttering analysis process is executed again when the trigger time is reached.
[0037] Step 103: If there are no abnormal CDN nodes in each CDN node, perform a second reset on the preset period, take the second reset period as the new preset period, and determine whether the new current time is the trigger time of the new preset period; the second reset period is greater than or equal to the preset period.
[0038] Specifically, if there are no abnormal CDN nodes among the current CDN nodes, it indicates that the overall CDN system is operating well and the service of each CDN node is normal, so there is no need to perform stuttering analysis at a high frequency. At this time, the preset period is reset for the second time, setting the preset period to 5 minutes. Since there are no abnormal CDN nodes at present, there is no need to frequently trigger the stuttering analysis process; performing stuttering analysis at a longer period can meet the monitoring requirements and reduce the overhead of system resources. After the second reset is completed, the current time continues to be monitored to determine whether the new current time is the trigger time for the new preset period (i.e., 5 minutes), so that the stuttering analysis process is executed again when the trigger time is reached.
[0039] The CDN-accelerated domain name lag analysis method provided in this application can proactively detect performance degradation of CDN node-accelerated domain names by automatically triggering the lag analysis process at a preset period. Compared with the existing technology of passively waiting for user complaints before investigation, it achieves proactive detection and timely handling of lag issues. By comprehensively judging three indicators—first byte time, interruption rate, and service bandwidth—abnormal CDN nodes are initially screened. Compared with the existing technology that relies on only a single indicator for anomaly judgment, it effectively reduces false positives and false negatives and avoids the computational overhead of performing full analysis of all CDN nodes. This system can acquire node operation status parameters, switch operation status parameters, host operation status parameters, and multi-dimensional traffic distribution parameters of abnormal CDN nodes in one step to perform stuttering analysis on accelerated domains. It can refine the anomaly location from the node level to the specific accelerated domain, overcoming the shortcomings of existing technologies that can only identify node anomalies but cannot locate specific domains. Moreover, it eliminates the need for manual cross-system data correlation analysis, improving troubleshooting efficiency. By dynamically adjusting the analysis cycle based on the presence of abnormal CDN nodes, it achieves an adaptive balance between domain stuttering analysis efficiency and system performance, thereby significantly improving the accuracy of CDN accelerated domain stuttering analysis.
[0040] In one embodiment, the step 101 mentioned, "determining whether there are abnormal CDN nodes in each CDN node based on the first byte time, interruption rate, and service bandwidth," is implemented in this embodiment by... Figure 2 Steps 201 and 202 shown are implemented.
[0041] Step 201: For each CDN node, determine whether the first byte time of the CDN node is higher than the first preset time threshold, whether the interruption rate of the CDN node is higher than the first interruption rate threshold, and whether the service bandwidth of the CDN node is higher than the first preset bandwidth threshold.
[0042] Step 202: If the first byte time is higher than the first preset time threshold, or the interruption rate is higher than the first preset interruption rate threshold, or the service bandwidth is higher than the first preset bandwidth threshold, then the CDN node is determined to be an abnormal CDN node.
[0043] Among them, CDN nodes whose service bandwidth is lower than the second preset bandwidth threshold are not identified as abnormal CDN nodes, and the first preset bandwidth threshold is greater than the second preset bandwidth threshold.
[0044] Specifically, from the CDN nodes at the current moment, normal CDN nodes with normal first byte time and interruption rate or excessively low business bandwidth are filtered out, and the remaining CDN nodes are marked as abnormal CDN nodes.
[0045] For example, it is determined whether the first byte time of each CDN node is higher than a first preset time threshold, whether the interruption rate is higher than a first interruption rate threshold, and whether the service bandwidth is higher than a first preset bandwidth threshold. If the first byte time of a CDN node is higher than the first preset time threshold, or its interruption rate is higher than the first interruption rate threshold, or its service bandwidth is higher than the first preset bandwidth threshold, then the CDN node is determined to be an abnormal CDN node. That is, if any one of the above three conditions is met, the CDN node is determined to be an abnormal CDN node.
[0046] It should be noted that during the judgment process, the following exclusion operation can also be performed: if the service bandwidth of a CDN node is lower than the second preset bandwidth threshold, then the CDN node will not be identified as an abnormal CDN node, regardless of whether its first byte time and interruption rate meet the above-mentioned abnormal conditions. The purpose of this exclusion operation is to filter out CDN nodes with extremely low service volume—because such CDN nodes carry very little service traffic, marking them as abnormal CDN nodes is likely to generate false alarms, and the actual value of analyzing the buffering of such CDN nodes with extremely low service volume is low, so there is no need to analyze such CDN nodes.
[0047] In one embodiment, the step 102 mentioned, "based on node operating status parameters, switch operating status parameters, host operating status parameters and / or multi-dimensional traffic distribution parameters, performing stuttering analysis on the accelerated domain names of abnormal CDN nodes to obtain the stuttering analysis results of the domain names at the current moment," in this embodiment, is achieved through... Figure 3 Steps 301 and 302 shown are implemented.
[0048] Step 301: If the retransmission rate of the abnormal CDN node is higher than the preset retransmission rate threshold, and the switch bandwidth of the abnormal CDN node is higher than the preset switch bandwidth threshold, and the switch health value of the CDN node is lower than the preset switch health threshold, then it is determined that the node running status parameters and the switch running status parameters meet the first judgment rule, and the domain name lag analysis result is determined to be a performance problem caused by the switch corresponding to the abnormal CDN node running at full capacity.
[0049] Step 302: If the node operating status parameters and switch operating status parameters do not meet the first judgment rule, and the node retransmission rate is higher than the preset node retransmission rate threshold, and the switch bandwidth is lower than the preset switch bandwidth threshold, and the switch health value is higher than the preset switch health threshold, then the node operating status parameters and switch operating status parameters meet the second judgment rule, and the domain name lag analysis result is determined to be a performance problem caused by hardware failure of the host group corresponding to the abnormal CDN node.
[0050] Among them, node retransmission rate is a node operating status parameter, while switch bandwidth and switch health value are switch operating status parameters.
[0051] Specifically, for any abnormal CDN node currently in operation, the first step is to determine if it meets the first judgment rule: if the abnormal CDN node's retransmission rate is higher than a preset retransmission rate threshold, and its switch bandwidth is higher than a preset switch bandwidth threshold, and its switch health value is lower than a preset switch health threshold, then the abnormal CDN node's node operating status parameters and switch operating status parameters meet the first judgment rule. If the first judgment rule is met, then the root cause of the accelerated domain name buffering issue is determined to be a performance problem caused by the switch bandwidth being fully utilized. That is, the switch is currently carrying traffic close to or has reached its maximum processing capacity, leading to increased packet transmission latency or even packet loss, thus causing the buffering phenomenon in the accelerated domain name buffering.
[0052] If the first judgment rule is not met, then the second judgment rule is continued to be judged: if the node retransmission rate of the abnormal CDN node is still higher than the preset node retransmission rate threshold, and the switch bandwidth of the abnormal CDN node is lower than the preset switch bandwidth threshold, and the switch health value of the abnormal CDN node is higher than the preset switch health threshold, then it is determined that the node operation status parameters and switch operation status parameters of the abnormal CDN node meet the second judgment rule.
[0053] When the second judgment rule is met, it is determined that the root cause of the lag in the accelerated domain name of the abnormal CDN node is a performance problem caused by hardware failure of the host group. That is, the root cause of the high retransmission rate of the node is not network congestion on the switch side (the switch bandwidth is not full and the health status is good), but a hardware failure at the backend host group level, such as host network card failure, memory error or disk damage, which leads to data transmission interruption or retransmission.
[0054] It should be noted that in the above judgment process, the first judgment rule has a priority matching order. First, it is judged whether the first judgment rule is satisfied. Only if the first judgment rule is not satisfied will it be judged whether the second judgment rule is satisfied. The two judgment rules will not be satisfied at the same time.
[0055] In one embodiment, the step 102 mentioned in "conducting a stuttering analysis on the accelerated domain name of the abnormal CDN node based on the node running status parameters, switch running status parameters, host running status parameters and / or multi-dimensional traffic distribution parameters, and obtaining the stuttering analysis result of the domain name at the current moment" can also be achieved through the following steps in this embodiment.
[0056] If the node operating status parameters and switch operating status parameters do not meet the second judgment rule, and the packet loss rate of the abnormal CDN node is higher than the preset node packet loss rate threshold, and the average service bandwidth of the abnormal CDN node is higher than the preset average bandwidth threshold, and the average health value of the abnormal CDN node is lower than the preset average health threshold, then the node operating status parameters and host operating status parameters meet the third judgment rule, and the domain name lag analysis result is determined to be a performance problem caused by the excessive load of the abnormal CDN node.
[0057] Among them, node packet loss rate is a node operating status parameter, while host average service bandwidth and host average health value are host operating status parameters.
[0058] Specifically, after obtaining the node running status parameters and host running status parameters of the abnormal CDN node, further analysis of the lag of the accelerated domain name of the abnormal CDN node can be performed based on these two types of parameters.
[0059] Node operational status parameters include node packet loss rate, and host operational status parameters include average host bandwidth and average host health value. The packet loss rate of the abnormal CDN node is compared with a preset node packet loss rate threshold; the average bandwidth of each host within the abnormal CDN node is compared with a preset host average bandwidth threshold; and the average health value of each host within the abnormal CDN node is compared with a preset host average health threshold. The average host bandwidth and average host health value are calculated as follows: the average host bandwidth is obtained by summing the bandwidth values of all hosts within the abnormal CDN node and dividing by the number of hosts; the average host health value is obtained by summing the health values of all hosts within the abnormal CDN node and dividing by the number of hosts.
[0060] The third judgment rule is determined as follows: If the packet loss rate of the abnormal CDN node is higher than the preset packet loss rate threshold, and the average service bandwidth of the host of the abnormal CDN node is higher than the preset average host bandwidth threshold, and the average host health value of the abnormal CDN node is lower than the preset average host health threshold, then the node operation status parameters and host operation status parameters of the abnormal CDN node are determined to meet the third judgment rule. When the third judgment rule is met, it is determined that the root cause of the acceleration domain name lag of the abnormal CDN node is a performance problem caused by excessive load. That is, the overall service traffic carried by each host within the abnormal CDN node is too large, approaching or exceeding the host's processing capacity limit, causing some requests to be unable to be processed in time and resulting in packet loss. The host health status also declines accordingly, thus causing the acceleration domain name lag.
[0061] It should be noted that the third judgment rule is only performed after the preceding judgment rules (the first and second judgment rules) are not met. That is, the combination of node retransmission rate and related switch parameters has already been assessed (corresponding to two root causes: switch bandwidth saturation and host group hardware failure). Only after ruling out problems on the switch side is the node packet loss rate, average host service bandwidth, and average host health value comprehensively assessed to determine if the performance issue is caused by excessive load. This judgment order reflects a layer-by-layer troubleshooting logic from the network link layer to the host processing layer.
[0062] In one embodiment, the step 102 mentioned in "conducting a stuttering analysis on the accelerated domain name of the abnormal CDN node based on the node running status parameters, switch running status parameters, host running status parameters and / or multi-dimensional traffic distribution parameters, and obtaining the stuttering analysis result of the domain name at the current moment" can also be achieved through the following steps in this embodiment.
[0063] If the node running status parameters and host running status parameters do not meet the third judgment rule, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average disk load of the abnormal CDN node is higher than the preset host average disk load threshold, then it is determined that the node running status parameters and host running status parameters meet the fourth judgment rule, and the domain name lag analysis result is determined to be a performance problem caused by the low hit rate of the abnormal CDN node.
[0064] Among them, node hit rate and node back-to-parent / back-to-source bandwidth are node operating status parameters, while host average disk load is a host operating status parameter.
[0065] Specifically, after obtaining the node running status parameters and host running status parameters of the abnormal CDN node, further analysis of the lag of the accelerated domain name of the abnormal CDN node can be performed based on these two types of parameters.
[0066] Node operational status parameters include node hit rate and node back-to-parent / back-to-origin bandwidth, while host operational status parameters include host average disk load. Back-to-parent bandwidth is the bandwidth consumption incurred by a CDN node when requesting data from its parent node (the next-level CDN node), and back-to-origin bandwidth is the bandwidth consumption incurred by a CDN node when requesting data from its origin server; both are collectively referred to as back-to-parent / back-to-origin bandwidth.
[0067] The hit rate of the abnormal CDN node is compared with a preset hit rate threshold. The back-to-parent / back-to-origin bandwidth of the abnormal CDN node is compared with a preset back-to-parent / back-to-origin bandwidth threshold. The average disk load of each host within the abnormal CDN node is compared with a preset average disk load threshold. The average disk load is calculated by summing the disk load values of all hosts within the abnormal CDN node and dividing by the number of hosts.
[0068] The fourth judgment rule is determined as follows: If the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node's back-to-parent / back-to-origin bandwidth is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host's average disk load is higher than the preset host's average disk load threshold, then the node's running status parameters and host's running status parameters are determined to meet the fourth judgment rule. When the fourth judgment rule is met, it is determined that the root cause of the accelerated domain name's lag due to the abnormal CDN node is a performance problem caused by a low hit rate. That is, the abnormal CDN node's cache hit rate is insufficient, and a large number of requests cannot be directly responded to from the cache, forcing them to retrieve data from the parent or origin, resulting in a surge in back-to-parent / back-to-origin bandwidth. At the same time, the data read pressure on the origin or parent node increases, the host disk reads and writes frequently, the disk load increases, the response speed decreases, and thus the lag phenomenon of the accelerated domain name occurs.
[0069] It should be noted that the fourth judgment rule is only determined after the preceding judgment rules (the first, second, and third judgment rules) have all failed. That is, only after ruling out three root causes—switch bandwidth saturation, host group hardware failure, and excessive load—is the system further assessed using node hit rate, node back-to-parent / back-to-source bandwidth, and average host disk load to determine if the performance issue is caused by a low hit rate. This judgment order reflects a step-by-step troubleshooting logic from the network link layer to the host processing layer and then to the caching service layer.
[0070] In one embodiment, the step 102 mentioned in "conducting a stuttering analysis on the accelerated domain name of the abnormal CDN node based on the node running status parameters, switch running status parameters, host running status parameters and / or multi-dimensional traffic distribution parameters, and obtaining the stuttering analysis result of the domain name at the current moment" can also be achieved through the following steps in this embodiment.
[0071] If the node running status parameters and host running status parameters do not meet the fourth judgment rule, and the traffic within a single client-domain node of the abnormal CDN node is higher than the preset traffic threshold or the number of requests per second is higher than the preset number of requests threshold, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the average host health value of the CDN node is lower than the preset average host health threshold, then the multidimensional traffic distribution parameters, node running status parameters, and host running status parameters meet the fifth judgment rule, and the domain name lag analysis result is determined to be a performance problem caused by a sudden increase in the client domain name traffic or the number of requests per second of the abnormal CDN node. Among them, the traffic within a single customer-domain node and the number of requests per second are multi-dimensional traffic distribution parameters, the node's back-to-parent / back-to-origin bandwidth and the node's hit rate are node operating status parameters, and the host's average health value is a host operating status parameter.
[0072] Multidimensional traffic distribution parameters include the traffic and requests per second of each accelerated domain name carried by the abnormal CDN node under different customers, different provinces and different operators.
[0073] Node status parameters include node hit rate and node back-to-parent / back-to-source bandwidth. Host status parameters include host average health value.
[0074] Traffic within a single customer-domain node refers to the total business traffic generated by a specific accelerated domain under a specific customer within the abnormal CDN node. Requests per second (PS) is the number of requests received per second by the specific accelerated domain under a specific customer within the abnormal CDN node. Node hit rate is the ratio of cache-hit requests within the abnormal CDN node to the total number of requests; a higher hit rate indicates better caching performance and fewer requests to the parent / origin server. Average host health is the average health score of all hosts within the abnormal CDN node; a higher health score indicates a more normal host operation, while a lower health score indicates potential performance degradation or failure risk.
[0075] Specifically, after obtaining the multi-dimensional traffic distribution parameters, node operation status parameters, and host operation status parameters of the abnormal CDN node, further congestion analysis can be performed on the accelerated domains of the abnormal CDN node based on these parameters. That is, it can be determined whether the traffic within a single client-domain node of the abnormal CDN node exceeds a preset traffic threshold or whether the number of requests per second exceeds a preset request threshold; whether the node hit rate of the abnormal CDN node is lower than a preset node hit rate threshold; whether the node's back-to-parent / back-to-origin bandwidth is higher than a preset node back-to-parent / back-to-origin bandwidth threshold; and whether the host's average health value of the abnormal CDN node is lower than a preset host average health threshold.
[0076] Determine if the fifth judgment rule is met: If the traffic within a single client-domain node of an abnormal CDN node is higher than the preset traffic threshold or the number of requests per second is higher than the preset number of requests threshold, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average health value of the abnormal CDN node is lower than the preset host average health threshold, then the multidimensional traffic distribution parameters, node operating status parameters, and host operating status parameters of the abnormal CDN node are determined to meet the fifth judgment rule.
[0077] When the fifth judgment rule is met, it is determined that the root cause of the lag in the accelerated domain of the abnormal CDN node is a performance problem caused by a sudden increase in the traffic or requests per second of the customer's domain. That is, the business traffic or number of requests for a specific accelerated domain of a specific customer in the abnormal CDN node increases sharply in a short period of time, which leads to a decrease in the cache hit rate of the abnormal CDN node. A large number of requests are forced to return to the parent or origin to obtain data, and the bandwidth for returning to the parent / origin surges accordingly. The host's health deteriorates due to the handling of the sudden increase in requests, thus causing the lag phenomenon of the accelerated domain.
[0078] It should be noted that the fifth judgment rule is performed only after the preceding judgment rules (the first, second, third, and fourth judgment rules) are all not met. That is, after ruling out the four root causes of full switch bandwidth, host group hardware failure, excessive load, and low hit rate, the system further judges whether the performance problem is caused by a sudden increase in traffic or requests per second within a single customer-domain node, combined with the node hit rate being lower than the threshold, the node's back-to-parent / back-to-origin bandwidth being higher than the threshold, and the host's average health value being lower than the threshold.
[0079] In addition, it should be noted that the fifth judgment rule can also be met as follows: a sudden increase in traffic or requests per second within a single client-domain node of an abnormal CDN node, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the average host health value of the CDN node is lower than the preset host average health threshold.
[0080] To determine whether a sudden increase in traffic or requests per second (PS) occurs within a single client-domain node of an abnormal CDN node, we can first query the log center for the traffic and QPS time-series curves of the client-domain carried by the abnormal CDN node in various provinces and operators over the past M minutes. Then, we perform data cleaning on the traffic and QPS time-series curves (e.g., removing outliers significantly exceeding 3 standard deviations or using median filtering for smoothing). We then use robust linear regression to fit the cleaned traffic and QPS time-series curves to obtain the current traffic slope (a_traffic) and QPS slope (a_qps). If a_traffic > 100 Mbps / minute and persists for more than two time windows, a traffic surge is determined; or, if a_qps > 5000 requests / minute, a QPS surge is determined.
[0081] In one embodiment, the step 102 mentioned, "based on node operating status parameters, switch operating status parameters, host operating status parameters and / or multi-dimensional traffic distribution parameters, performing stuttering analysis on the accelerated domain name of the abnormal CDN node to obtain the stuttering analysis result of the domain name at the current moment," can also be achieved through... Figure 4 Steps 401 and 402 shown are implemented.
[0082] Step 401: If the multidimensional traffic distribution parameters, node running status parameters, and host running status parameters do not meet the fifth judgment rule, and the first byte time of a single host of the abnormal CDN node is higher than the second preset time threshold, and the single host interruption rate is higher than the second preset interruption rate threshold, and the single host health value is lower than the preset host health threshold, and the single host disk load is higher than the preset host disk load threshold, then it is determined that the host running status parameters meet the sixth judgment rule, and the domain name lag analysis result is determined to be that the single host disk abnormality of the abnormal CDN node caused the node performance problem.
[0083] Step 402: If the host operating status parameters do not meet the sixth judgment rule, and the first byte time of a single host is higher than the second preset time threshold, and the interruption rate of a single host is higher than the second preset interruption rate threshold, and the health value of a single host is lower than the preset host health threshold, and the CPU load of a single host is higher than the preset CPU load threshold, and the download speed of a single host is lower than the preset download speed threshold, then it is determined that the host operating status parameters meet the seventh judgment rule, and the domain name lag analysis result is determined to be a single host anomaly of an abnormal CDN node causing node performance problems.
[0084] Step 403: If the host's operating status parameters do not meet the seventh judgment rule, then the domain name lag analysis result is determined to be an unknown performance problem.
[0085] Among them, the single-host first byte time, single-host interruption rate, single-host health value, single-host disk load, single-host CPU load, and single-host download speed are host operating status parameters.
[0086] The single-host first-byte time is the time taken for a host within the abnormal CDN node to process a user request, from receiving the request to returning the first byte of response data; the single-host interruption rate is the ratio of the number of requests interrupted or failed by the host per unit time to the total number of requests received by the host; the single-host health value is a score obtained after quantitatively evaluating the overall operating status of the host; the single-host disk load is the disk I / O utilization of the host, that is, the percentage of disk read / write operations to its maximum processing capacity; the single-host CPU load is the central processing unit utilization of the host, that is, the percentage of CPU processing tasks to its maximum processing capacity; the single-host download rate is the speed at which the host transmits data to the user terminal.
[0087] Specifically, the above parameters are compared with preset thresholds. That is, it is determined whether the sixth judgment rule is met: if the single-host first byte time of the abnormal CDN node is higher than the second preset time threshold, and the single-host interruption rate of the abnormal CDN node is higher than the second preset interruption rate threshold, and the single-host health value of the abnormal CDN node is lower than the preset host health threshold, and the single-host disk load of the abnormal CDN node is higher than the preset host disk load threshold, then it is determined that the host operating status parameters of the abnormal CDN node meet the sixth judgment rule.
[0088] When the sixth judgment rule is met, it is determined that the root cause of the lag in the accelerated domain name of the abnormal CDN node is a single host disk abnormality causing node performance problems. That is, the disk load of a certain host in the abnormal CDN node is too high, which causes the data read and write speed of the host to drop significantly and the response latency to requests to increase significantly. At the same time, the health score of the host is lower than the normal level, indicating that the disk may be about to fail or has already failed, thus affecting the access experience of the accelerated domain name service on the host.
[0089] Conversely, if none of the above four conditions are met simultaneously, i.e., the sixth judgment rule is not met, then the seventh judgment rule is continued to be judged: if the first byte time of the abnormal CDN node's single host is still higher than the second preset time threshold, and the interruption rate of the abnormal CDN node's single host is still higher than the second preset interruption rate threshold, and the health value of the abnormal CDN node's single host is still lower than the preset host health threshold, and the CPU load of the abnormal CDN node's single host is higher than the preset CPU load threshold, and the download rate of the abnormal CDN node's single host is lower than the preset download rate threshold, then it is determined that the host operating status parameters of the abnormal CDN node meet the seventh judgment rule.
[0090] When the seventh judgment rule is met, it is determined that the root cause of the slowdown in the accelerated domain name of the abnormal CDN node is a single host malfunction causing node performance problems. That is, the CPU load of a certain host within the abnormal CDN node has approached or reached its processing capacity limit, while the download speed of this host is still too low, the response time is longer, and combined with the extended first byte time, increased interruption rate, and decreased health value, it indicates that the overall operating status of this host is abnormal, possibly caused by CPU resource exhaustion leading to comprehensive performance degradation. If the seventh judgment rule is also not met, then the acceleration domain name lag analysis result of the abnormal CDN node is determined to be an unknown performance problem. That is, the currently available host running status parameters cannot be attributed to the known lag root cause type, and further manual investigation is required.
[0091] It should be noted that the judgment in this embodiment is made under the premise that the preceding judgment rules (first judgment rule, second judgment rule, third judgment rule, fourth judgment rule, and fifth judgment rule) are not met. That is, after successively eliminating five root causes such as switch bandwidth saturation, host group hardware failure, excessive load, low hit rate, and sudden increase in traffic to specific customer domain names, the performance problem is further determined by various parameters at the single host level to determine whether it is caused by a single host disk anomaly or a comprehensive single host anomaly. The sixth and seventh judgment rules are both analyzed at the single host level, rather than at the entire node level, so that the location of the root cause of the lag can be further refined from the node level to the specific host level within the node.
[0092] In one embodiment, considering the convenience of subsequent queries and tracing, the results of each determined domain name lag analysis and its related parameters can be stored together. Therefore, after step 103, this embodiment can also complete the storage through the following steps.
[0093] Based on node operating status parameters, switch operating status parameters, host operating status parameters, multi-dimensional traffic distribution parameters, and domain name congestion analysis results, generate and store the domain name congestion analysis record corresponding to the current moment.
[0094] Specifically, after obtaining the domain name congestion analysis results for the abnormal CDN nodes at the current moment, the node operating status parameters, switch operating status parameters, host operating status parameters, and multi-dimensional traffic distribution parameters obtained during this congestion analysis, along with the final determined domain name congestion analysis results, are used to generate a domain name congestion analysis record corresponding to the current moment. This domain name congestion analysis record uses the current timestamp as the primary key identifier to ensure that each generated domain name congestion analysis record has a unique timestamp.
[0095] Domain name lag analysis records are stored using a many-to-one association structure of master and child records. In one embodiment, the content of the master record includes, but is not limited to: the current timestamp (i.e., the current moment when the current lag analysis process is triggered), the node name of the abnormal CDN node, the node operating status parameters of the abnormal CDN node (e.g., node retransmission rate, node packet loss rate, node hit rate, node back-to-parent / back-to-origin bandwidth, node download speed, etc.), the switch operating status parameters of the switch corresponding to the abnormal CDN node (e.g., switch bandwidth, switch health value, etc.), the host operating status parameters of each host within the abnormal CDN node (host network card bandwidth, host service bandwidth, host back-to-parent / back-to-origin bandwidth, host hit rate, host packet loss rate, host retransmission rate, host download speed, host first byte time, host interruption rate, host CPU load, host disk load, host health value, etc.), and the domain name lag analysis results determined in this analysis (e.g., "performance problems caused by switch running at full capacity", "performance problems caused by excessive load", "performance problems caused by low hit rate", "performance problems caused by sudden increase in customer domain name traffic or requests per second", "node performance problems caused by single host disk anomaly", "node performance problems caused by single host anomaly", or "unknown performance problems").
[0096] The content of the sub-record is derived from the multi-dimensional traffic distribution parameters of the abnormal CDN node. From these parameters, the traffic and requests per second (RPS) of each accelerated domain name carried by the abnormal CDN node under different customers, provinces, and operators are extracted. A separate distribution sub-record is generated for each dimension combination using "customer-domain-province-operator". Each sub-record includes: customer identifier, accelerated domain name, province, operator, traffic value for that dimension combination, and RPS for that dimension combination.
[0097] Filter out distribution sub-records with excessively low traffic or requests per second (e.g., traffic below a preset filtering traffic threshold and / or requests per second below a preset filtering request threshold), and associate each of the remaining distribution sub-records with the main record of this domain congestion analysis in a many-to-one relationship, all linked to the same current timestamp. Then, store the main record and its associated distribution sub-records together in the database.
[0098] Through the aforementioned many-to-one association storage structure of master records and distributed sub-records, operations and maintenance personnel can subsequently filter queries based on conditions such as time range, node name, customer identifier, accelerated domain name, province, and ISP to analyze the causes of domain name lag. For example, a filtering query statement can be generated based on conditions such as time range, node name, customer identifier, accelerated domain name, province, and ISP. This query statement can then be associated with the master record and distributed sub-records in the database before being filtered and the results returned. This allows for quick identification of the cause of domain name lag under specific conditions.
[0099] It should be noted that for the node operation status parameters, switch operation status parameters, and host operation status parameters of abnormal CDN nodes, other indicators not included in this domain name congestion analysis can be used for subsequent viewing or manual analysis by operations and maintenance personnel (host network card bandwidth, host service bandwidth, host parent / origin bandwidth, host hit rate, host packet loss rate, host retransmission rate, host download speed, host first byte time, host interruption rate, host CPU load, host disk load, host health value, etc.). In one embodiment, considering that the accumulated amount of domain name lag analysis records increases over time, failure to clean them up in a timely manner will consume a large amount of storage resources and affect subsequent retrieval efficiency. Therefore, this embodiment completes the periodic cleaning of domain name lag records through the following steps.
[0100] Perform a domain lag analysis record clearing operation at fixed intervals, deleting all domain lag analysis records from before the preset time period.
[0101] Specifically, the system performs a buffering analysis record clearing operation at fixed intervals. This fixed interval can be pre-configured by the system, for example, it can be set to 1 hour, meaning the system performs the domain buffering analysis record clearing operation every hour. When performing the domain buffering analysis record clearing operation, the system deletes all domain buffering analysis records created before a preset time period. The preset time period can be pre-configured, for example, it can be set to 30 days, meaning that each time the domain buffering analysis record clearing operation is performed, all domain buffering analysis records created more than 30 days ago (including the main record and its associated distributed sub-records) are deleted. In this way, the system can control the total number of domain buffering analysis records in the database within a reasonable range, maintaining sufficient database storage space and good query performance.
[0102] For example, refer to Figure 5 The diagram shown illustrates the process of analyzing CDN-accelerated domain name buffering. Figure 5 In this context, the node granularity index is specifically the node operating status parameter in the aforementioned embodiments, the switch granularity index is the switch operating status parameter in the aforementioned embodiments, the host granularity index is the host operating status parameter in the aforementioned embodiments, and the traffic distribution granularity index is specifically the multidimensional traffic distribution parameter in the aforementioned embodiments.
[0103] The buffering analysis process is initiated periodically according to a preset cycle. Upon each initiation, the system first queries metrics such as first-byte time, interruption rate, and service bandwidth of each CDN node. Based on this, abnormal CDN nodes exhibiting performance degradation are identified, while CDN nodes with excessively low service bandwidth are filtered out to avoid invalid analysis. If no abnormal CDN nodes are found, the process waits for the next cycle to restart. If abnormal CDN nodes are found, the system further queries node-level metrics, switch-level metrics, host-level metrics, and traffic distribution-level metrics of those abnormal CDN nodes. By comprehensively analyzing these multi-level metrics, the root cause of the buffering is determined, and a domain buffering analysis record is generated and stored in the database.
[0104] In addition, a historical lag record cleanup operation is performed periodically as an unscheduled task, deleting historical domain lag analysis records older than a specified number of days to control database storage capacity. Users can use the interface to flexibly filter and query stored domain lag analysis records based on conditions such as time range, node, customer, domain name, province, and ISP to quickly locate the cause of domain lag in specific scenarios. The specific process involved can be referred to the aforementioned embodiment. It will not be repeated here.
[0105] Reference Figure 6 The diagram shown is a structural block diagram of a CDN-accelerated domain name buffering analysis system. Figure 6 The CDN acceleration domain name lag analysis system includes an indicator collection module, a lag analysis module, a record storage module, a query interface module, an information display module, and a record cleanup module.
[0106] The metrics acquisition module connects to the monitoring center and log center of the CDN system. It collects metrics such as first byte time, interruption rate and service bandwidth of each CDN node from the monitoring center. After identifying abnormal CDN nodes, it further collects node-level metrics, switch-level metrics, host-level metrics and traffic distribution-level metrics of the abnormal CDN node to provide data support for the lag analysis module.
[0107] The lag analysis module starts periodically according to a preset cycle. Based on the first byte time, interruption rate and service bandwidth provided by the indicator collection module, it filters out abnormal CDN nodes. When there are abnormal CDN nodes, it analyzes multiple levels of indicators such as node granularity indicators, switch granularity indicators, host granularity indicators and traffic distribution granularity indicators to find the root cause of the lag. It generates a domain name lag analysis record containing the current timestamp, various indicator data and lag cause analysis results, and stores the domain name lag analysis record in the record storage module.
[0108] The record storage module stores domain name lag analysis records generated by the lag analysis module for retrieval by the query interface module. The query interface module provides an Application Programming Interface (API) that supports filtering and querying stored lag records based on conditions such as time range, node, customer, domain name, province, and ISP. The information display module provides a web-based interactive query page based on the query interface, divided into a lag record list page and a record details page, to present the query results to the user. The record details page can display all the indicator parameters included in node granularity indicators, switch granularity indicators, host granularity indicators, and traffic distribution granularity indicators. The record cleanup module deletes historical domain name lag analysis records older than a specified number of days (e.g., 30 days) from the record storage module at fixed intervals (e.g., hourly) to control storage capacity. The specific processes involved can be referred to in the aforementioned embodiments and will not be repeated here.
[0109] based on Figures 6-7 This application has the following core advantages: First, by jointly analyzing data from the monitoring center and log center, node and host anomalies are linked to accelerated domain name services, thereby improving the accuracy of CDN domain name lag analysis.
[0110] Secondly, the system automatically diagnoses and analyzes and generates structured logs of lag, avoiding repetitive manual analysis by maintenance personnel and effectively reducing the cost of manual troubleshooting.
[0111] Third, it provides filtering queries based on time range, node, customer, domain name, province, operator, etc., to more efficiently locate the cause of domain name lag.
[0112] It should be noted that although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.
[0113] The following is for reference. Figure 7 It shows a schematic diagram of the structure of a computer system 500 suitable for implementing terminal devices or servers in the embodiments of this application.
[0114] like Figure 7As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 502 or programs loaded from storage section 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the system 500. The CPU 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.
[0115] The following components are connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 510 as needed so that computer programs read from it can be installed into storage section 508 as needed.
[0116] In particular, according to embodiments of this disclosure, the above references Figure 1 The described process can be implemented as a computer software program. For example, embodiments of this disclosure include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program containing instructions for performing... Figure 1 The program code for the method. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable media 511.
[0117] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.
[0118] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0119] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be located in a processor. The names of these units or modules do not, in certain circumstances, constitute a limitation on the unit or module itself.
[0120] On the other hand, this application also provides a computer-readable storage medium, which may be included in the computer device described in the above embodiments, or may exist independently and not assembled into the computer device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the methods described in this application. For example, it may execute... Figure 1 The steps of the method shown are as follows.
[0121] This application provides a computer program product including instructions that, when executed, cause the method described in this application to be performed. For example, it can execute... Figure 1 The steps of the method shown are as follows.
[0122] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0123] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.
Claims
1. A method for analyzing CDN-accelerated domain name buffering issues, characterized in that, The method includes: If the current time is the trigger time of the preset period, the first byte time, interruption rate and service bandwidth of each CDN node are obtained, and based on the first byte time, interruption rate and service bandwidth, it is determined whether there are abnormal CDN nodes among the CDN nodes. In the case where an abnormal CDN node exists among the CDN nodes, the node operating status parameters, switch operating status parameters, host operating status parameters, and multi-dimensional traffic distribution parameters of the abnormal CDN node are obtained. Based on the node operating status parameters, switch operating status parameters, host operating status parameters, and / or the multi-dimensional traffic distribution parameters, a stuttering analysis is performed on the accelerated domain name of the abnormal CDN node to obtain the stuttering analysis result of the domain name corresponding to the current time. Then, the preset period is reset for the first time, and the period after the first reset is used as the new preset period. It is then determined whether the new current time is the trigger time of the new preset period; the period after the first reset is less than the preset period. If no abnormal CDN node is found among the CDN nodes, the preset period is reset for the second time, and the period after the second reset is used as the new preset period. It is then determined whether the new current time is the trigger time of the new preset period. The period after the second reset is greater than or equal to the preset period.
2. The method according to claim 1, characterized in that, The step of determining whether there are abnormal CDN nodes among the CDN nodes based on the first byte time, the interruption rate, and the service bandwidth includes: For each CDN node, determine whether the first byte time of the CDN node is higher than a first preset time threshold, whether the interruption rate of the CDN node is higher than a first interruption rate threshold, and whether the service bandwidth of the CDN node is higher than a first preset bandwidth threshold. If the first byte time is higher than the first preset time threshold, or the interruption rate is higher than the first preset interruption rate threshold, or the service bandwidth is higher than the first preset bandwidth threshold, then the CDN node is determined to be the abnormal CDN node. Among them, CDN nodes whose service bandwidth is lower than the second preset bandwidth threshold are not identified as abnormal CDN nodes, and the first preset bandwidth threshold is greater than the second preset bandwidth threshold.
3. The method according to claim 1, characterized in that, The step involves performing a stuttering analysis on the accelerated domain name of the abnormal CDN node based on the node's operating status parameters, the switch's operating status parameters, the host's operating status parameters, and / or the multi-dimensional traffic distribution parameters, to obtain the stuttering analysis result for the domain name at the current moment, including: If the retransmission rate of the abnormal CDN node is higher than the preset retransmission rate threshold, and the switch bandwidth of the abnormal CDN node is higher than the preset switch bandwidth threshold, and the switch health value of the CDN node is lower than the preset switch health threshold, then it is determined that the node running status parameters and the switch running status parameters meet the first judgment rule, and it is determined that the domain name lag analysis result is that the performance problem is caused by the switch corresponding to the abnormal CDN node running at full capacity. If the node operating status parameters and the switch operating status parameters do not meet the first judgment rule, and the node retransmission rate is higher than the preset node retransmission rate threshold, and the switch bandwidth is lower than the preset switch bandwidth threshold, and the switch health value is higher than the preset switch health threshold, then it is determined that the node operating status parameters and the switch operating status parameters meet the second judgment rule, and it is determined that the domain name lag analysis result is a performance problem caused by hardware failure of the host group corresponding to the abnormal CDN node. The node retransmission rate is a node operating status parameter, and the switch bandwidth and the switch health value are switch operating status parameters.
4. The method according to claim 3, characterized in that, The method further includes: If the node operating status parameters and the switch operating status parameters do not meet the second judgment rule, and the node packet loss rate of the abnormal CDN node is higher than the preset node packet loss rate threshold, and the host average service bandwidth of the abnormal CDN node is higher than the preset host average bandwidth threshold, and the host average health value of the abnormal CDN node is lower than the preset host average health threshold, then it is determined that the node operating status parameters and the host operating status parameters meet the third judgment rule, and it is determined that the domain name lag analysis result is that the abnormal CDN node is overloaded, causing performance problems. The node packet loss rate is a node operating status parameter, and the host average service bandwidth and the host average health value are host operating status parameters.
5. The method according to claim 4, characterized in that, The method further includes: If the node running status parameters and the host running status parameters do not meet the third judgment rule, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average disk load of the abnormal CDN node is higher than the preset host average disk load threshold, then it is determined that the node running status parameters and the host running status parameters meet the fourth judgment rule, and it is determined that the domain name lag analysis result is that the low hit rate of the abnormal CDN node caused the performance problem. Among them, the node hit rate and the node back-to-parent / back-to-source bandwidth are node operating status parameters, and the host average disk load is host operating status parameter.
6. The method according to claim 5, characterized in that, The method further includes: If the node running status parameters and the host running status parameters do not meet the fourth judgment rule, and the traffic within a single client-domain node of the abnormal CDN node is higher than the preset traffic threshold or the number of requests per second is higher than the preset number of requests threshold, and the node hit rate of the abnormal CDN node is lower than the preset node hit rate threshold, and the node back-to-parent / back-to-origin bandwidth of the abnormal CDN node is higher than the preset node back-to-parent / back-to-origin bandwidth threshold, and the host average health value of the CDN node is lower than the preset host average health threshold, then it is determined that the multidimensional traffic distribution parameters, the node running status parameters, and the host running status parameters meet the fifth judgment rule, and it is determined that the domain name lag analysis result is a performance problem caused by a sudden increase in the client domain name traffic or the number of requests per second of the abnormal CDN node. Among them, the traffic within a single client-domain node and the number of requests per second belong to the multidimensional traffic distribution parameters, the node back-to-parent / back-to-origin bandwidth and the node hit rate belong to the node operating status parameters, and the host average health value belongs to the host operating status parameters.
7. The method according to claim 6, characterized in that, The method further includes: If the multidimensional traffic distribution parameters, the node running status parameters, and the host running status parameters do not meet the fifth judgment rule, and the single host first byte time of the abnormal CDN node is higher than the second preset time threshold, the single host interruption rate is higher than the second preset interruption rate threshold, the single host health value is lower than the preset host health threshold, and the single host disk load is higher than the preset host disk load threshold, then it is determined that the host running status parameters meet the sixth judgment rule, and it is determined that the domain name lag analysis result is that the single host disk abnormality of the abnormal CDN node caused the node performance problem; If the host operating status parameters do not meet the sixth judgment rule, and the first byte time of a single host is higher than the second preset time threshold, and the single host interruption rate is higher than the second preset interruption rate threshold, and the single host health value is lower than the preset host health threshold, and the single host CPU load is higher than the preset CPU load threshold, and the single host download speed is lower than the preset download speed threshold, then it is determined that the host operating status parameters meet the seventh judgment rule, and it is determined that the domain name lag analysis result is a node performance problem caused by the single host anomaly of the abnormal CDN node; If the host operating status parameters do not meet the seventh judgment rule, then the domain name lag analysis result is determined to be an unknown performance problem; Among them, the single-host first byte time, the single-host interruption rate, the single-host health value, the single-host disk load, the single-host CPU load, and the single-host download speed are all host operating status parameters.
8. The method according to claim 1, characterized in that, The method further includes: Based on the node operating status parameters, the switch operating status parameters, the host operating status parameters, the multidimensional traffic distribution parameters, and the domain name lag analysis results, a domain name lag analysis record corresponding to the current moment is generated and stored.
9. The method according to claim 8, characterized in that, The method further includes: Perform a domain lag analysis record clearing operation at fixed intervals, deleting all domain lag analysis records from before the preset time period.
10. A computer program product, characterized in that, The computer program product includes instructions that, when executed, cause the method as described in any one of claims 1-9 to be implemented.
