In some embodiments, the risk information includes a risk level and a risk rate. The operation S220 may specifically include the following sub-operations S21 to S24.

At sub-operation S21, a time period used when the risk level is obtained by calculation is taken as a calculation-involved time period.

At sub-operation S22, a first-state duration proportion in each calculation-involved time period is obtained according to a ratio of the first state duration in each calculation-involved time period to the corresponding calculation-involved time period.

At sub-operation S23, normalization processing is performed based on the number of state changes in each calculation-involved time period and a corresponding state change number threshold, as well as the first-state duration proportion in each calculation-involved time period and a corresponding first-state duration proportion threshold, so as to obtain risk rates of the link in each calculation-involved time period.

At sub-operation S24, a maximum rate value of the risk rates of the link in the respective calculation-involved time periods is taken as the risk rate of the link.

The risk rate of each link in the network can be obtained by calculation through the above sub-operations S21 to S24.

In some embodiments, the normalization processing in the operation S23 may include, for example, the following sub-operations S31 to S33.

At sub-operation S31, first normalization processing is performed on the number of state changes in the calculation-involved time period and the corresponding state change number threshold to obtain the risk rate of the link caused by a state change in the calculation-involved time period.

In an example, in a case where the first state is the link-up state, the first normalization processing may be performed according to the following formula (1-1) to obtain the risk rate caused by the state change in the calculation-involved time period.

f
downNum

(

downNum
,

threshold
⁢
1

)

=

k
⁢
1

+

k
⁢
2

1
+

e

1
-

downNum

threshold
⁢
1

(

1
-
1

)

In the formula (1-1), f_downNumrepresents the risk rate caused by the number of state changes of the link from the link-up state to the link-down state, and downNum is the number of state changes of a current link from the link-up state to the link-down state in the calculation-involved time period, and threshold1 is the state change number threshold corresponding to the calculation-involved time period. Illustratively, k1=0.4 and k2=0.4. It should be understood that the values of k1 and k2 may be customized according to actual situations, and are not specifically limited in the embodiments of the present disclosure.

In an example, in a case where the first state is the link-down state, the first normalization processing may be performed according to the following formula (1-2) to obtain the risk rate caused by the state change in the calculation-involved time period.

f
upNum

(

upNum
,

threshold
⁢
1

)

=

k
⁢
1

+

k
⁢
2

1
+

e

1
-

upNum

threshold
⁢
1

(

1
-
2

)

In the formula (1-2), f_upNumrepresents the risk rate caused by the number of state changes of the link from the link-down state to the link-up state, and upNum is the number of state changes of a current link from the link-down state to the link-up state in the calculation-involved time period, and the meaning of threshold1 and the values of k1 and k2 are the same as those in the formula (1-1), and thus will not be repeated here.

At sub-operation S32, second normalization processing is performed based on the first-state duration proportion in the calculation-involved time period and the corresponding first-state duration proportion threshold to obtain the risk rate of the link caused by the first-state duration proportion in the calculation-involved time period.

In an example, in the case where the first state is the link-up state, the second normalization processing may be performed according to the following formula (2-1) to obtain the risk rate caused by the operation duration proportion in the calculation-involved time period.

f
validRadio
up

(

uptime
,

threshold
⁢
2

)

=

k
⁢
3

-

uptime

threshold
⁢
2

×
k
⁢
4

(

2
-
1

)

In the formula (2-1),

f
validRadio
up

is the risk rate caused by the operation duration proportion, uptime is the operation duration proportion in the calculation-involved time period, and threshold2 is the operation duration proportion threshold corresponding to the calculation-involved time period. Illustratively, k3=0.8 and k4=0.2. It should be understood that the values of k3 and k4 may be customized according to actual situations, and are not specifically limited in the embodiments of the present disclosure.

In an example, in the case where the first state is the link-down state, the second normalization processing may be performed according to the following formula (2-2) to obtain a first-state idle duration proportion involved in the calculation and a corresponding first-state duration proportion threshold.

f
validRadio
down

(

downtime
,

threshold
⁢
2

)

=

k
⁢
3

-

1
-
downtime

threshold
⁢
2

×
k
⁢
4

(

2
-
2

)

In the formula (2-2),

f
validRadio
down

is the risk rate caused by the idle duration proportion, downtime is the idle duration proportion in the calculation-involved time period, and the meaning of threshold2 and the values of k3 and k4 are the same as those in the formula (2-1), and thus will not be repeated here.

At sub-operation S33, the maximum rate value of the risk rates is acquired from the risk rate caused by the state change and the risk rate caused by the first-state duration proportion and taken as the risk rate corresponding to the risk level of the link.

In an example, if the middle time period is taken as the calculation-involved time period, the risk rate caused by the state change for the middle time period may be expressed as the following formula (3):

f
downNum
middle

=

f
downNum

(

downnum
downNum
middle

,

threshold
downNum
middle

)

(
3
)

In the formula (3),

f
downNum
middle

is the risk rate caused by the number of state changes corresponding to the calculation-involved middle time period,

downnum
downNum
middle

is the number of state changes corresponding to the calculation-involved middle time period, and

threshold
downNum
middle

is the state change number threshold corresponding to the calculation-involved middle time period.

In this example, for the middle time period, if the first state is the link-up state, the first-state duration proportion is the operation duration proportion, and the risk rate caused by the operation duration proportion may be expressed as the following formula (4):

f
validRadio
middle

=

f
validRadio
up

(

uptime
validRadio
middle

,

threshold
validRadio
middle

)

(
4
)

In the formula (4),

f
validRadio
middle

is the risk rate caused by the operation duration proportion corresponding to the calculation-involved middle time period,

uptime
validRadio
mddle

is the operation duration proportion corresponding to the calculation-involved middle time period, and

threshold
validRadio
middle

is the operation duration proportion threshold corresponding to the calculation-involved middle time period.

In an example, if the longest time period is taken as the calculation-involved time period, the risk rate caused by the state change in the longest time period may be expressed as the following formula (5):

f
downNum
long

=

f
downNum

(

downnum
downNum
long

,

threshold
downNum
long

)

(
5
)

In the formula (5), f_downNum^longis the risk rate caused by the number of state changes corresponding to the calculation-involved longest time period,

downnum
downNum
long

is the number of state changes corresponding to the calculation-involved longest time period, and

threshold
downNum
long

is the state change number threshold corresponding to the calculation-involved longest time period.

In this example, for the longest time period, if the first state is the link-up state, the risk rate caused by the change of the first state may be expressed as the following formula (6):

f
validRadio
long

=

f
validRadio
up

(

uptime
validRadio
long

,

threshold
validRadio
long

)

(
6
)

In the formula (6),

f
validRadio
long

is the risk rate caused by the operation duration proportion corresponding to the calculation-involved longest time period,

uptime
validRadio
long

is the operation duration proportion corresponding to the calculation-involved longest time period, and

threshold
validRadio
long

is the operation duration proportion threshold corresponding to the calculation-involved longest time period.

In some embodiments, the first and second normalization processings described above may also be normalization methods based on z-score or normalization methods based on Hyperbolic Tangent Function (Tanth).

In an example, the z-score based normalization methods may be expressed as the following formula (7):

f
⁢
1
⁢

(
x
)

=

(

x
-
μ

)

σ

(
7
)

In the formula (7), if the value of x is the number of state changes from the first state to the second state in the calculation-involved time period, μ is an average of the numbers of state changes from the first state to the second state in the preset N time periods, σ is a variance of the numbers of state changes from the first state to the second state in the preset N time periods, and f1(x) is the risk rate obtained by calculation and caused by the state change corresponding to the calculation-involved time period; and if the value of x is the first-state duration proportion in the calculation-involved time period, μ is an average of the first-state duration in the preset N time periods, σ is a variance of the first-state duration in the preset N time periods, and f1(x) is the risk rate obtained by calculation and caused by the first-state duration proportion corresponding to the calculation-involved time period.

In an example, the Tanth-based normalization methods may be expressed as the following formula (8):

f
⁢
2
⁢

(
x
)

=

e
x

-

e

-
x

e
x

+

e

-
x

(
8
)

In the formula (8), if the value of x is the number of state changes from the first state to the second state in the calculation-involved time period, f2 (x) is the risk rate obtained by calculation and caused by the state change corresponding to the calculated-involved time period; and if the value of x is the first-state duration proportion in the calculation-involved time period, f2(x) is the risk rate obtained by calculation and caused by the first-state duration proportion corresponding to the calculation-involved time period.

It should be understood that the above formulae of the first normalization processing and the second normalization processing may also be formulae of other normalization processing methods, may be specifically set according to actual application scenarios, and are not specifically limited in the embodiments of the present disclosure.

In some embodiments, the risk information includes a risk level and a risk rate. The operation S230 may specifically include the following operations S41 and S42.

At operation S41, the risk level of the link is added to topology map information of a network to which the link belongs to obtain updated topology map information, and the path computation on a preset service is triggered according to the updated topology map information.

At operation S42, during the path computation: for the link with a high risk level, the link with the high risk level is set as a disconnected link; for the link with a low risk level, the path computation is still performed; and for the link with a medium risk level, the path computation is performed based on a risk rate of the link with the medium risk level.

Through the above operations, when the preset service is rerouted, the risk information of each link can be added to map resources needed by the path computation. In this way, the risk suppression in the operation S240 can make use of different risk levels to avoid the risk links as much as possible, so as to achieve the purpose of suppressing the risk links. Thus, the path computation can be performed based on the risk information of the links: setting the high risk link as the disconnected link, taking the low risk link as the link which does not affect rerouting computation, and adding a corresponding risk rate strategy to the path computation of the medium risk link, so that the risk information and the risk rates of the links can be considered during the path computation, and risk avoidance and risk suppression can be carried out according to corresponding results of the path computation.

In some embodiments, the risk suppression comprises rerouting suppression. The operation of performing the path computation based on the risk rate of the link with the medium risk level in the operation S42 may specifically include the following operations S42-01 to S42-03.

At operation S42-01, the path computation is performed on the preset service using a preset path computation strategy to obtain a candidate rerouting path of the preset service; and the preset path computation strategy includes: an original rerouting computation strategy and a newly added risk rate strategy which is a computation strategy configured according to the risk rate of the link with the medium risk level.

At operation S42-02, the sum of risk rates of all links in each candidate rerouting path is calculated to obtain an accumulated risk rate of each candidate rerouting path.

At operation S42-03, the candidate rerouting path corresponding to a minimum rate value of the accumulated risk rates is selected as an optimal switch path of the path computation.

Through the above operations, in the process of performing the path computation using the path computation strategy, a plurality of path computation strategies may be adopted, the plurality of path computation strategies include the original rerouting computation strategy and the newly added risk rate strategy; and among the results of the path computation combined with the newly added risk rate strategy, the path with the minimum accumulated risk rate is selected as the optimal switch path of the rerouting, thereby achieving risk avoidance on the path switching of the rerouting of the service.

In some embodiments, performing the path computation on the preset service using the preset path computation strategy includes: performing the path computation on the preset service according to priority levels of the strategies included in the preset path computation strategy; and among the strategies included in the preset path computation strategy, a priority level of the risk rate strategy is lower than that of a highest priority strategy that is used for preferentially completing the rerouting of the preset service.

In the present embodiment, the original rerouting computation strategy may include, for example, a strategy needed by the rerouting, such as a minimal-hop strategy, and the risk rate strategy is the strategy of lowest priority level, that is, a link with the lowest accumulated risk rate is selected as far as possible on the basis that other strategies are satisfied; considering that the rerouting needs to be completed preferentially, the risk rate strategy may also adopt, for example, a second-lowest priority strategy, that is, the link with the lowest accumulated risk rate is preferentially selected on the basis that the rerouting strategy is completed preferentially.

In some embodiments, the risk suppression includes switch-back suppression. The operation of performing the path computation based on the risk rate of the link with the medium risk level in the operation S42 may specifically include the following operations S42-04 and S42-05.

At operation S42-04, the sum of risk rates of all links in the current service path of the preset service is calculated and taken as a current-path risk rate, and the sum of risk rates of all links in an original service path of the preset service is calculated and taken as an original-path risk rate.

At operation S42-05, in a case where the original-path risk rate is less than the current-path risk rate, the original-path risk rate is taken as a result of the path computation of the switch-back of the service.

Through the above operations, when the path of the service needs to be switched back, the accumulated risk rate of all the links in the current route and the accumulated risk rate of the switch-back route are calculated, and the path is switched back when the accumulated risk rate of the switch-back route is smaller; and the path is not switched back otherwise. Thus, risk avoidance on the path switching of the rerouting of the service is realized.

In some embodiments, the link in the embodiments of the present disclosure is a link composed of underlying network devices connected to a network controller via a southbound interface; the network controller is a network controller to which the link processing apparatus of the embodiment of the present disclosure belongs; and the state of the link is a link state information derived in advance from the received link change messages of the link, and the link change messages are messages reported by the underlying network devices via the southbound interface.

The link processing method according to an embodiment of the present disclosure further includes: sending the result of the path computation to the underlying network devices via the southbound interface of the network controller, so as to cause the underlying network devices to perform the path switching according to the result of the path computation.

In this embodiment, the network controller issues the result of the path computation to the device via the southbound interface for completion of the path switching.

In some embodiments, for the link with the high risk level, the link processing method further includes: in a case where the path computation is performed again, setting a penalty weight for the link with the high risk level, so as to reduce, according to the penalty weight, the number of services deployed on the link with the high risk level.

In this embodiment, for the link with the high risk level, the penalty weight for the link may be added during the subsequent path computation, so that the number of the services on the link can be reduced, and finally, frequent switching of the services can be adaptively suppressed from a management and control level, and invalid path switching caused by oscillating links can be effectively suppressed.

In some embodiments, the risk information includes a risk rate; and the link control method further includes: determining a link having a risk rate greater than a preset risk rate threshold in a network as a link prone to oscillation; and generating corresponding risk prompt information for the determined link prone to oscillation.

In the embodiments of the present disclosure, in view of the problem existing in the link oscillation, the link processing method according to the embodiment of the present disclosure can effectively identify the risk links having different influences, and can effectively suppress invalid rerouting and invalid switch-back caused by an oscillating link by adopting cross-level countermeasures; and the link processing method provided by the embodiments of the present disclosure can further provide the risk prompt for the link which may possibly oscillate in the future, and add the penalty weight to the link during the subsequent path computation, thereby reducing the number of the services on the link, and finally adaptively suppressing the frequent switching of the services from the management and control level.

FIG. 3 is a detailed flowchart illustrating a link processing method according to an exemplary embodiment of the present disclosure. As shown in FIG. 3, the link processing method may include the following operations S301 to S314.

At operation S301, data updating of a link is performed.

In this operation, data updating of links is classified into two modes: one is data updating of link state changes of all links triggered by a timer, and the other is data updating of a link state change of any link performed in response to link state change information which is reported by the any link and received via the southbound interface of the controller. The subsequent processes corresponding to the two data updating modes are the same.

At operation S302, parameter initialization is performed.

In this operation, a state change number threshold and an operation duration proportion threshold of the link may be set for each preset time period.

At operation S303, the numbers of state changes and the operation duration proportions of the link in different preset time periods are counted.

Illustratively, the different preset time periods include the shortest time period, the middle time period, and the longest time period that are described in the above embodiments.

At operation S304, it is determined whether the number of state changes of the link in the shortest time period exceeds the state change number threshold of the link; if the number of state changes in the shortest time period exceeds the state change number threshold, operation S305 is performed to determine the risk level of the link as high risk; and if the number of state changes in the shortest time period does not exceed the state change number threshold, operation S306 is performed.

At operation S306, it is determined whether the number of state changes in a non-shortest time period exceeds the state change number threshold corresponding to the non-shortest time period, or whether the operation duration proportion in the non-shortest time period exceeds the operation duration proportion threshold corresponding to the non-shortest time period. If the number of state changes in the non-shortest time period exceeds the state change number threshold corresponding to the non-shortest time period, or the operation duration proportion in the non-shortest time period exceeds the operation duration proportion threshold corresponding to the non-shortest time period, operation S307 is performed to determine the risk level of the link as medium risk; and if the number of state changes in the non-shortest time period does not exceed the state change number threshold corresponding to the non-shortest time period, or the operation duration proportion in the non-shortest time period does not exceed the operation duration proportion threshold corresponding to the non-shortest time period, operation S308 is performed to determine the risk level of the link as low risk.

Through the operations S301 to S308, risk identification of the link is completed.

At operation S309, map resources are updated.

In this operation, the risk information is added to the map resources needed by the path computation. The link with the high risk level is set as the disconnected link, the path computation is still performed on the link the low risk level, and the risk rate strategy is added to the link with the medium risk level.

At operation S310, as shown by “multi-strategy path computation” in FIG. 3, in a rerouting stage with risk suppression, a plurality of path computation strategies are adopted to perform the path computation. The plurality of path computation strategies include the original rerouting computation strategy and the newly added risk rate strategy.

At operation S311, in the result of the path computation, the link with the lowest accumulated risk rate is selected as the optimal path to complete the rerouting.

At operation S312, as shown by “the risk rate of the current route is smaller than that of the switch-back route” in FIG. 3, in a switch-back stage with risk suppression, it is determined whether the accumulated risk rate of the current route obtained by the computation is smaller than the accumulated risk rate of the switch-back route obtained by the computation; if the accumulated risk rate of the current route obtained by the computation is not smaller than the accumulated risk rate of the switch-back route obtained by the computation, which indicates that the accumulated risk rate of the switch-back route is smaller, operation S313 is performed to determine that switch-back of the link is to be performed; and if the accumulated risk rate of the current route obtained by the computation is smaller than the accumulated risk rate of the switch-back route obtained by the computation, operation S314 is performed to determine that the switch-back of the link is not to be performed.

Through the operations S306 to S311, the risk suppression for the link in the rerouting and switch-back processes is realized.

In the link processing method according to the embodiment of the present disclosure, the process of identifying the risk links is based on the real historical data of the links, each link in the network is independently analyzed, the algorithms are clear and do not belong to black box algorithms, an identification rate of the risk links can be more than 90%, and a high risk-identification rate can be achieved.

Furthermore, according to the embodiment of the present disclosure, prediction of a future oscillation situation of the links is achieved based on the counting of the historical data of the links. Since the time taken by the algorithms of risk identification is very short, the time consumption of the algorithms can be neglected in management and control of products, which allows for rapid risk identification.

In the embodiments of the present disclosure, the algorithms of risk identification and the rerouting/switch-back path computation can be combined, the risk rate strategy can be embedded in the path computation algorithms, and the invalid rerouting and switch-back caused by the oscillation can be effectively and adaptively suppressed. For a link which oscillates with high frequency, the link can be suppressed a hundred percent; and for a link which oscillates with low frequency, the link can be efficiently suppressed on the premise of ensuring that the network is actually unobstructed, thereby realizing a high suppression rate of the risk links.

FIG. 4 is a schematic diagram of a system architecture according to another embodiment of the present disclosure. The structures in FIG. 4 which are the same as or equivalent to the structures in FIG. 1 are given the same reference numerals. As shown in FIG. 4, the system architecture shown in FIG. 4 is different from that shown in FIG. 1 in that the network controller 10 in the system architecture in FIG. 4 may include: a risk identification module 101 and a risk suppression module 102.

The risk identification module 101 may be configured to calculate a risk rate of a link; and the risk suppression module 102 may be configured to perform path computation based on risk information of the link, so as to perform risk suppression on path switching according to a result of the path computation.

The risk identification module 101 is capable of timely responding to state change information of the link uploaded by each network device in the network 20, and is capable of determining and outputting the risk information of the link in real time.

The risk information in the embodiments of the present disclosure may include: a risk level and a risk rate. The risk level (i.e., a risk degree) may include, for example, three levels, i.e., low risk, medium risk, and high risk, and each risk level has a corresponding risk rate with a value range of 0 to 1. Specifically, for the high risk, the corresponding risk rate obtained by calculation may be, for example, greater than or equal to 0.8; for the medium risk, the corresponding risk rate obtained by calculation may be, for example, greater than or equal to 0.6 and less than 0.8; and for the low risk, the corresponding risk rate obtained by calculation may be, for example, less than 0.6.

In the risk suppression module 102, for a link identified as a high-risk link, the link is suppressed regardless of a state of the link, that is, computation for rerouting or switch-back is not performed until the link is not a high risk link any more. For a medium-risk link, computation for rerouting or switch-back is still performed, but is performed with an influence of the risk rate taken into consideration. For a low-risk link, computation for rerouting or switch-back is not affected.

In this system architecture, the network controller 10 may trigger risk link identification according to a received link oscillation message (e.g., a link change message reported by a network device via the southbound interface module 11), and the risk identification module 101 starts to calculate a risk level and a risk rate of a currently triggered link; the risk suppression module 102 starts to perform rerouting or switch-back, and the risk rate is preferentially considered in the path computation during the rerouting or switch-back; and after the risk suppression module 102 completes the path computation, the risk suppression module 102 feeds the result back to the network controller 10. The network controller 10 issues the result of the path computation to the network device via the southbound interface module 11 for completion of the path switching.

A link processing apparatus according to the embodiments of the present disclosure is described in detail below with reference to the drawings. FIG. 5 is a schematic structural diagram of a link processing apparatus according to an embodiment of the present disclosure. As shown in FIG. 5, the link processing apparatus may include the following modules.

An information acquisition module 510 is configured to acquire the number of state changes and first state duration of a link in a preset time period.

A risk calculation module 520 is configured to calculate risk information of the link based on the number of state changes and the first state duration.

A path computation module 530 is configured to perform path computation based on the risk information.

A risk suppression module 540 is configured to perform risk suppression on path switching according to a result of the path computation.

In some embodiments, the risk information includes a risk level; and the number of the preset time period is at least one. The risk calculation module 520 includes: a high-risk determination unit configured to determine a risk level of the link as high risk if it is determined that the number of state changes in a shortest time period is greater than or equal to a change number threshold corresponding to the shortest time period; and the shortest time period is a minimum value among the at least one preset time period, and is less than or equal to a preset time period threshold. The risk calculation module 520 further includes a medium-risk determination unit configured to determine, among the at least one preset time period, a risk level of the link in an acquired non-shortest time period as medium risk if it is determined that the number of state changes in the acquired non-shortest time period meets a corresponding change number limit condition or it is determined that the first state duration in the acquired non-shortest time period meets a corresponding duration limit condition. The risk calculation module 520 further includes a low-risk determination unit configured to determine the risk level of the link in the acquired non-shortest time period as low risk if it is determined that the number of state changes in the acquired non-shortest time period does not meet the corresponding change number limit condition and the first state duration in the acquired non-shortest time period does not meet the corresponding duration limit condition.

In some embodiments, the change number limit condition is: the number of state changes in the acquired non-shortest time period is greater than or equal to a preset change number threshold corresponding to the acquired non-shortest time period. In a case where a first state is a link-up state and a second state is a link-down state, the first state duration is operation duration, and the duration limit condition is: an operation duration proportion of the link in the non-shortest time period is less than an operation duration proportion threshold corresponding to the acquired non-shortest time period.

In some embodiments, in a case where the first state is the link-down state and the second state is the link-up state, the first state duration is idle duration, and the duration limit condition is: an idle duration proportion of the link in the acquired non-shortest time period is greater than or equal to an idle duration proportion threshold corresponding to the acquired non-shortest time period.

In some embodiments, the risk information includes a risk level and a risk rate. The risk calculation module 520 includes: a time-period determination unit configured to take a time period used when a risk level is obtained by calculation as a calculation-involved time period; a duration proportion calculation unit configured to obtain a first-state duration proportion in each calculation-involved time period according to a ratio of the first state duration in each calculation-involved time period to the corresponding calculation-involved time period; a risk rate calculation unit configured to perform normalization processing based on the number of state changes in each calculation-involved time period and a corresponding state change number threshold, and the first-state duration proportion in each calculation-involved time period and a corresponding first-state duration proportion threshold, so as to obtain risk rates of the link in each calculation-involved time period; and a maximum-risk-rate determination unit configured to take a maximum rate value of the risk rates of the link in each calculation-involved time period as a risk rate of the link.

In some embodiments, when performing the normalization processing, the risk rate calculation unit is further configured to: perform first normalization processing on the number of state changes in the calculation-involved time period and the corresponding state change number threshold to obtain the risk rate of the link caused by a state change in the calculation-involved time period, perform second normalization processing based on the first-state duration proportion in the calculation-involved time period and the corresponding first-state duration proportion threshold to obtain the risk rate of the link caused by the first-state duration proportion in the calculation-involved time period, and acquire the maximum rate value of the risk rates from the risk rate caused by the state change and the risk rate caused by the first-state duration proportion and take the maximum rate value of the risk rates as the risk rate corresponding to the risk level of the link.