Service abnormity detection method and server

Abstract

The invention provides a service abnormity detection method and a management server. The method comprises following steps that: the management server calculates to obtain dynamic baseline values of service indexes and determines the normal interval of the service indexes; then the service index data of present time are obtained; if it is determined that the service index data are located out of the normal interval of the service indexes, warning information is generated. According to the method and the server of the invention, the normal interval of the service indexes is determined through the calculated dynamic baseline values, especially using an average value and a variance; whether the service index data are abnormal is determined; and the measurability and the maintainability of the service system are improved.

Description

technical field [0001] The invention relates to the field of information technology, in particular to a method and server for detecting business abnormality. Background technique [0002] At present, in the monitoring of network business systems, the general method is that each network unit of the business system periodically reports some business index data (including device status data) to the management server, and the management server then confirms according to some configured business index thresholds. Whether there is an abnormality in the business or equipment, so as to perform operations such as alarms. For example, the service network element reports data such as new user visits, network access traffic, and CPU usage to the management server. The management server determines whether to send an alarm according to the alarm threshold configured by the administrator. For example, when the network access traffic reaches 10Mb / s, an alarm is generated, or when the CPU ...

AI Technical Summary

Problems solved by technology

For example, when the network access traffic reaches 10Mb / s, an alarm is generated, or when the CPU usage reaches 85%, an alarm is generated, etc., and data such as the number of new user visits is usually used for statistics, because this cannot directly set an alarm threshold.

[0003] This kind of alarm method in which different business indicators depend on specific values ​​is mainly for equipment processing capabilities and equipment status, and does not make a judgment on whether the business itself is normal. For example, the busyness of business is usually different at different times of the day. At night, there is usually less business, and the CPU usage rate reaching 50% may be abnormal caused by some software failure, while the CPU usage rate is lower than 20% during the day, which may be caused by business configuration and other factors.

None of the existing technologies can detect such anomalies

Images