Fault detection method, apparatus, and system
In certain embodiments, a forwarding device receives at least one service flow. The forwarding device obtains service information of the at least one service flow, where the service information of the service flow includes identification information of a network object to which the service flow belongs and M key performance indicators KPIs of the service flow. M is an integer greater than 0, and the network object includes one or more devices. The forwarding device sends detection information to a first device, where the detection information includes the service information of the at least one service flow or a feature set obtained based on the service information of the at least one service flow. The detection information is used to detect whether the network object is in a faulty state.
1. A method, comprising:
receiving, by a forwarding device, a service flow;
obtaining, by the forwarding device, service information of the service flow, wherein the service information of the service flow comprises identification information of a network object to which the service flow belongs and M key performance indicators (KPIs) of the service flow, M is an integer greater than 0, and the network object comprises a device,
wherein the M KPIs comprises, based on a target service packet from the service flow further based on configuration policy information:
a network delay between the forwarding device and the network object to which the service flow belongs,
a first data volume that is of the service flow and sent by the network object,
a second data volume that is of the service flow and received by the network object, or
a status identifier that identifies a status of the service flow; and
sending, by the forwarding device, detection information to a first device to detect whether the network object is in a faulty state, wherein the detection information comprises the service information of the service flow or a feature set obtained based on the service information of the service flow.
2. The method according to claim 1 , wherein:
a protocol type of the service flow is Transmission Control Protocol (TCP); and
the method further comprises:
obtaining, by the forwarding device, the target service packet from the service flow based on the configuration policy information, wherein the configuration policy information comprises a preset packet type; and
obtaining, by the forwarding device, the M KPIs of the service flow based on the target service packet.
3. The method according to claim 2 , wherein:
obtaining, by the forwarding device, the M KPIs of the service flow based on the target service packet comprises:
obtaining, by the forwarding device in response to the target service packet comprising a first target service packet and a second target service packet, the network delay between the forwarding device and the network object based on a first time at which the first target service packet is received and a second time at which the second target service packet is received, wherein the first target service packet is a packet sent to the network object, and the second target service packet is a packet that is sent by the network object and that corresponds to the first target service packet; or
obtaining, by the forwarding device in response to the target service packet comprising a first start packet and a first end packet, and based on a sequence number of the first start packet and a sequence number of the first end packet, the first data volume that is of the service flow and sent by the network object, wherein the first start packet is a first packet of the service flow and sent by the network object, and the first end packet is a last packet of the service flow and sent by the network object; or
obtaining, by the forwarding device in response to the target service packet comprising a second start packet and a second end packet, and based on a sequence number of the second start packet and a sequence number of the second end packet, the second data volume that is of the service flow and received by the network object, wherein the second start packet is a first packet of the service flow and received by the network object, and the second end packet is a last packet of the service flow and received by the network object.
4. The method according to claim 2 , wherein:
the method further comprises setting, by the forwarding device in response to the target service packet comprising a first start packet, in a first time length after a third time, and according to whether the forwarding device receives a first end packet, the status identified by the status identifier to a success state or a failure state; and
the third time is a time at which the first start packet is received, the first start packet is a first packet that is of the service flow and sent by the network object, and the first end packet is a last packet that is of the service flow and sent by the network object.
5. The method according to claim 1 , wherein the method further comprises, before sending, by the forwarding device, the detection information to the first device:
obtaining, by the forwarding device from the service flow, KPIs of N service flows that are in a current cycle and that belong to a target network object, wherein the target network object is a network object to which the service flow belongs, and N is an integer greater than 0; and
obtaining, by the forwarding device, a feature set based on the KPIs of the N service flows.
6. The method according to claim 5 , wherein:
the feature set comprises at least one statistical feature; and
obtaining, by the forwarding device, the feature set based on the KPIs of the N service flows comprises:
obtaining, by the forwarding device, M KPI sets, wherein any KPI set of the M KPI sets comprises one KPI of each of the N service flows, and types of the KPIs of the any KPI set are the same; and
performing, by the forwarding device in at least one first calculation manner, calculation on the KPIs of the any KPI set, to obtain at least one statistical feature corresponding to the any KPI set, wherein the at least one first calculation manner comprises one or more of:
collecting statistics on the KPIs of the any KPI set; or
calculating an average value, a variance, dispersion, skewness, or kurtosis of the KPIs of the any KPI set.
7. The method according to claim 6 , wherein:
the feature set further comprises at least one time domain feature; and
the method further comprises, performing, by the forwarding device after performing the calculation on the KPIs of the any KPI set, calculation on statistical features of the statistical feature set in at least one second calculation manner, to obtain at least one time domain feature; and
the statistical feature set comprises K statistical features, the K statistical features are statistical features of a same type and obtained through calculation in K cycles, the K cycles comprise the current cycle and K−1 cycles before the current cycle, and the at least one second calculation manner comprises one or more of:
calculating a period-on-period value or a differential value between two adjacent statistical features in the statistical feature set; or
performing feature fitting on the statistical features in the statistical feature set.
8. The method according to claim 6 , wherein:
the any KPI set comprises status identifiers of the N service flows, the status identifier of any one of the N service flows identifies a status of the any service flow, the statistical feature of the any KPI set comprises a quantity of status identifiers that identify a success state and a quantity of status identifiers that identify a failure state, and the feature set further comprises a proportion of service flows in the success state or a proportion of service flows in the failure state; and
the method further comprises, after performing the calculation on the KPIs of the any KPI set:
calculating the proportion of the service flows in the success state based on the quantity of status identifiers that identify the success state and a quantity of KPIs of the any KPI set; or
calculating the proportion of the service flows in the failure state based on the quantity of status identifiers that identify the failure state and the quantity of KPIs of the any KPI set.
9. A method, comprising:
receiving, by a first device, service information of a service flow and sent by a forwarding device, wherein the service information of the service flow comprises identification information of a network object to which the service flow belongs and M key performance indicators (KPIs) of the service flow, M is an integer greater than 0, and the network object comprises a device,
wherein the M KPIs comprises, based on a target service packet from the service flow further based on configuration policy information:
a network delay between the forwarding device and the network object to which the service flow belongs,
a first data volume that is of the service flow and sent by the network object,
a second data volume that is of the service flow and received by the network object, or
a status identifier that identifies a status of the service flow; and
detecting, by the first device based on the service information of the service flow using a fault detection model, whether the network object is in a faulty state, or obtaining, based on the service information of the service flow, at least one feature set for detecting the network object.
10. The method according to claim 9 , wherein detecting, by the first device based on the service information of the service flow using the fault detection model, whether the network object is in the faulty state comprises:
obtaining, by the first device from the service flow, KPIs of N service flows that are in a current cycle and that belong to a target network object, wherein the target network object is a network object to which the service flow belongs, and N is an integer greater than 0;
obtaining, by the first device, a feature set based on the KPIs of the N service flows; and
detecting, by the first device based on the feature set using the fault detection model, whether the target network object is in the faulty state.
11. The method according to claim 10 , wherein:
the feature set comprises at least one statistical feature; and
obtaining, by the first device, the feature set based on the KPIs of the N service flows comprises obtaining, by the first device, M KPI sets, wherein any KPI set comprises one KPI of each of the N service flows, and types of the KPIs of the any KPI set are the same; and
performing, by the first device, calculation on the KPIs of the any KPI set in at least one first calculation manner, to obtain at least one statistical feature corresponding to the any KPI set, wherein the at least one first calculation manner comprises one or more of:
collecting statistics about the KPIs in the any KPI set; or
calculating an average value, a variance, dispersion, skewness, or kurtosis of the KPIs of the any KPI set.
12. The method according to claim 11 , wherein:
the feature set further comprises at least one time domain feature; and
the method further comprises, after performing the calculation on the KPIs of the any KPI set: performing calculation on statistical features of the statistical feature set in at least one second calculation manner, to obtain at least one time domain feature, wherein the statistical feature set comprises K statistical features, the K statistical features are statistical features of a same type and obtained through calculation in K cycles, the K cycles comprise the current cycle and K−1 cycles before the current cycle, and the at least one second calculation manner comprises one or more of:
calculating a period-on-period value or a differential value between two adjacent statistical features in the statistical feature set; or
performing feature fitting on the statistical features in the statistical feature set.
13. The method according to claim 9 , further comprising, in response to detecting that a target network object is in the faulty state:
obtaining, by the first device, a KPI of the target network object or a service flow of the target network object; and
performing fault locating based on the KPI of the target network object or the service flow of the target network object.
14. An apparatus, comprising:
at least one processor; and
at least one non-transitory computer-readable storage medium storing a program executable by the at least one processor, the program comprising instructions to:
receive a service flow;
obtain service information of the service flow, wherein the service information of the service flow comprises identification information of a network object to which the service flow belongs and M key performance indicators (KPIs) of the service flow, M is an integer greater than 0, and the network object comprises a device,
wherein the M KPIs comprises, based on a target service packet from the service flow further based on configuration policy information:
a network delay between the apparatus and the network object to which the service flow belongs,
a first data volume that is of the service flow and sent by the network object,
a second data volume that is of the service flow and received by the network object, or
a status identifier that identifies a status of the service flow; and
send detection information to a first device to detect whether the network object is in a faulty state, wherein the detection information comprises the service information of the service flow or a feature set obtained based on the service information of the service flow.
15. The apparatus according to claim 14 , wherein:
a protocol type of the service flow is Transmission Control Protocol (TCP); and
the program further includes instructions to:
obtain the target service packet from the service flow based on the configuration policy information, wherein the configuration policy information comprises a preset packet type; and
obtain the M KPIs of the service flow based on the target service packet.
16. The apparatus according to claim 15 , wherein:
the program further includes instructions to:
obtain, in response to the target service packet comprising a first target service packet and a second target service packet, the network delay between the apparatus and the network object based on a first time at which the first target service packet is received and a second time at which the second target service packet is received, wherein the first target service packet is a packet sent to the network object, and the second target service packet is a packet that is sent by the network object and that corresponds to the first target service packet; or
obtain, in response to the target service packet comprising a first start packet and a first end packet, and based on a sequence number of the first start packet and a sequence number of the first end packet, the first data volume that is of the service flow and sent by the network object, wherein the first start packet is a first packet of the service flow and sent by the network object, and the first end packet is a last packet of the service flow and sent by the network object; or
obtain, in response to the target service packet comprising a second start packet and a second end packet, and based on a sequence number of the second start packet and a sequence number of the second end packet, the second data volume that is of the service flow and received by the network object, wherein the second start packet is a first packet of the service flow and received by the network object, and the second end packet is a last packet of the service flow and received by the network object.
17. The apparatus according to claim 15 , wherein:
wherein the program further includes instructions to, when the target service packet comprises a first start packet, in a first time length after a third time:
set, if a first end packet is received, the status identified by the status identifier to a success state; or
set, if the first end packet is not received, the status identified by the state identifier to a failure state; and
the third time is a time at which the first start packet is received, the first start packet is a first packet that is of the service flow and sent by the network object, and the first end packet is a last packet that is of the service flow and sent by the network object.
18. The apparatus according to claim 14 , wherein the program further includes instructions to:
obtain, from the service flow, KPIs of N service flows that are in a current cycle and that belong to a target network object, wherein the target network object is a network object to which the service flow belongs, and N is an integer greater than 0; and
obtain a feature set based on the KPIs of the N service flows.
19. The apparatus according to claim 18 , wherein:
the feature set comprises at least one statistical feature; and
the program further includes instructions to:
obtain M KPI sets, wherein any KPI set of the M KPI sets comprises one KPI of each of the N service flows, and types of the KPIs of the any KPI set are the same; and
perform calculation on the KPIs of the any KPI set in at least one first calculation manner, to obtain at least one statistical feature corresponding to the any KPI set, wherein the at least one first calculation manner comprises one or more of:
collecting statistics about the KPIs of the any KPI set; or
calculating an average value, a variance, dispersion, skewness, or kurtosis of the KPIs of the any KPI set.
20. The apparatus according to claim 19 , wherein:
the feature set further comprises at least one time domain feature; and
the program further includes instructions to perform calculation on statistical features of the statistical feature set in at least one second calculation manner, to obtain at least one time domain feature;
the statistical feature set comprises K statistical features, the K statistical features are statistical features of a same type and obtained through calculation in K cycles, the K cycles comprise the current cycle and K−1 cycles before the current cycle, and the at least one second calculation manner comprises one or more of:
calculating a period-on-period value or a differential value between two adjacent statistical features in the statistical feature set; and
performing feature fitting on the statistical features in the statistical feature set.