网络安全检测|网络安全服务|网络安全扫描-香港墨客投资移动版

The work of Chu et al. [20] is based on the destination of each flow on the fault link, so each of the nodes on the forwarding path of the flow should be configured in advance for any possi- ble link failure on the forwarding path. In order to avoid link over- head, some works [31–33] try to use the SDN switches to manage the traffic in the network. Other works [21,34] based on the full topology to achieve load balancing for the links at the exit ports of the SDN switches. All these schemes must collect the workload on each link and the traffic volume between all source-destination pairs by the controller when the link fails, so the recovery time will be larger than 50ms.

Cascone et al. [35] deploy SDN switches in enterprise networks using VLANs and switch ports to enable each flow to go through at least one SDN switch. The main contribution of et al. [36] is to systematically tackle the problem of how to partially deploy SDN-enabled switches into existing legacy networks, and reap the benefits of the controllable network, subject to budget and re- source constraints. However, the scheme is limited by the number of VLANs, and it also needs to set the switch ports.

Caria et al. [34] focus on the hybrid SDN network scenario and discuss about the migration algorithms from the perspective of traffic engineering in an ISP network. It proposes a new architec- ture for a hybrid SDN/OSPF operation, in which a few SDN node is used to partition the OSPF domain into multiple sub-domains. Caria et al. [34] mainly concentrates on searching for an optimized migration sequence of the routers in the traditional IP network to minimize the maximum link utilization. But the SDN switches de- ployment scheme is not able to work for some topologies. The traf- fic scheduling in the intra-domain cannot be optimized. Cascone et al. [35] present two traffic management applications that exploit a stateful data plane and their prototype implementation based on OpenState, an OpenFlow evolution that recently proposed. The au- thors present two applications, namely forwarding consistency and failure recovery, that greatly benefit from the stateful SDN data plane. However, it will not be able to achieve traffic load balanc- ing. The program requires to pre-configure a large number of flow table entries, which occupy a large number of ternary content ad- dressable memory (TCAM) resources in SDN switches, and these flow table entries will not be used for a very long time until the distant link fails. In our scheme, we can effectively reduce and bal- ance the flow table entries of different switches even though many new flows arrive in the network at the same time.

7 Conclusions

In this paper, we formulate the problem of reducing and bal- ancing the flow table entries to hold more flows with low overhead as an optimization problem. The problem is proved to be NP-hard. Hence, we propose the greedy algorithm KSGT to solve the prob- lem. Compared with the existing solutions, KSGT can reduce about 60% of flow entries when processing the same amount of flows; and improve the successful installation and forwarding flows by about 25% under the same flow table space in SDN switches. Con- sidering the limited flow table space in SDN switches and MPLS labels overhead in the network, KSGT is the best choice to be used in real networks. We present our future work in two aspects. Theo- retically, on the one hand, we will try to improve the classification efficiency of the flows in order to better schedule the flows. On the other hand, we need to pay attention to the Multi-level flowtable structure and analyze the flow table aggregation algorithm based on TCAMs.

(责编：温静、赵光霞)