| As the development of network technology, the traditional carrier-class network gradually exposes the limitations of the data transfer. Ethernet shows an unprecedented activity because of it's simple standard and lower cost. But the Ethernet previously was only applied to local area networks, so it is an important thing for Ethernet to provide comprehensive OAM (operation management and maintenance) for MAN and WAN level carrier-class network. Ethernet ring network is the normal form of carrier-class network. Ring protection switching and fault recovery is an important component of Metro Ethernet OAM. G.8032 standard is developed by ITU. It can achieve the requirement that the recovery time is no more than 50ms and the user does not perceive the recovery process. So it is the most dominant and promising standard among the Metro Ethernet ring protection standards.G.8032 standard is improved in this paper. In order to solve the state of affairs that "message to allow open" (NRRB) sent by main node is lost, a recover timer is designed in this paper. The timer ensures that the failed link which has been restored could transmit data when "message to allow open" (NRRB) sent by main node is lost. In view that Ethernet ring network according to the existing protocol will repeatly count link failure times, a delay timer whose value is equal to the largest network latency of message(latency is milliseconds) is designed. Master node will start the delay timer on receipt of the first fault message. While the timer is running, master node will not receive any other fault message. Because Ethernet ring recovery scheme for two points failure is not considered comprehensively in G.8032, when two failures occur in the ring, some nodes in the ring could not communicate properly and the original protocol can not open the link that has been restored in time. This leads to the case that the link restored physically is still blocked in function. To avoid such situation, a solution based on recover timer to open the first restored link is provided in this paper. In addition to above, a protection and restore suggestion for the situation that there is a failure at both ends of the two inter-nodes is given in this paper. It overcomes the G.8032's weakness, and enhances the reliability of the Ethernet ring.The protection and restoration system based on single Ethernet ring is coded. The designed software is divided to several modules, which are initialization module, timer module, receiver and sender module, state machine module, and other major modules. In the initialization module, structures and variables that will be used during the implementation are assigned, the initial state of nodes is set and some preparatory work is completed. The start and stop methods of the timer is designed in the timer module. The send and receive treatment of kinds of messages is designed in receiver and sender module. Finally, the state machine module is responsible for action conversion.At the end of the paper, the function of the designed software is tested. The test contents include capabilities of failover and recovery and the recovery of multi-point failure in case of single-loop. The test results show that the designed software meets the basic requirements. The time between failure and switching to the standby is 40ms, which is less than 50ms. |
PDF Full Text Request