A Method of Predicting the Optimal Timeout Time of Flow Entry

Abstract

The invention discloses a method for forecasting optimal timeout of a flow table item. The method runs in a software-defined network system structure based on the OpenFlow. The method includes the steps of firstly, forecasting the throughput of the data flow and balancing resource use data of an exchanger flow table of the OpenFlow; secondly, forecasting the optimal timeout of the flow table item according to the resource use data through an optimized work flow algorithm. According to the method, performance of different services is improved by continuously optimizing timeout of the flow table item; when flash crowd occurs in a network, the method can effectively relieve the flash crowd, an exchanger can be forwarded to be used, and therefore the lose rate of the flash crowd is decreased, and the capacity, for forwarding data packets, of the exchanger is improved. By forecasting the optimal timeout of the flow table item, forwarding stability and forwarding usability of data in an SDN are effectively improved.

Description

technical field [0001] The invention relates to the technical field of software-defined networks, in particular to a method for predicting the optimal timeout time of a flow entry. Background technique [0002] Software-defined networking (SDN) separates the integrated control plane and data plane in traditional networks, enhancing the flexibility of network management and control. The SDN controller on the control plane manages the switch on the data plane through the control channel, and the switch processes the data flow according to the forwarding rules defined in the flow table configured by the controller. However, a problem that cannot be ignored in the SDN network architecture is: there is a performance bottleneck between the switch and the controller. When the software-defined network is applied to the WAN, it will face the problem of burst flow, and a large number of flows will cause the switch flow table entry If the number exceeds the upper limit, the control fl...

AI Technical Summary

Problems solved by technology

However, a problem that cannot be ignored in the SDN network architecture is: there is a performance bottleneck between the switch and the controller. When the software-defined network is applied to the WAN, it will face the problem of burst flow, and a large number of flows will cause the switch flow table entry If the number exceeds the upper limit, the control flow is congested, thus affecting the forwarding performance

[0003] It is more complicated to implement this method under the separated network architecture of SDN, because this method is based on the resource situation of the switch, and the bandwidth of the control channel between the controller and the switch is limited.

Images