Fault recovery and resource allocation in software-defined optical networks

Preface

Traditional IP packet switching networks use the Interior Gateway Protocol (IGP) and Border Gateway Protocol (BGP) to achieve universal interconnection and build a large-scale Internet on a global scale. However, with the rapid development of mobile Internet and the Internet of Things, new services emerge in an endless stream, and business needs are also changing. Traditional distributed networks are overwhelmed. At the same time, due to the characteristics of the control plane and the data plane being integrated into one, traditional networks also face many problems such as network congestion, complex equipment, difficult operation and maintenance, and slow deployment of new services. Software Defined Networking (SDN) breaks the vertical integration of traditional networks, and realizes the characteristics of centralized control logic and open network programming interfaces by separating control and forwarding, injecting new vitality into the network and allowing network managers to flexibly configure and reconfigure the network.

[[208333]]

The programmability and openness of SDN also provide new opportunities for the innovation of optical networks. After experiencing time division multiplexing (TDM) and wavelength division multiplexing (WDM), optical networks are also moving towards the future of flexible elastic optical networks (EONs) based on orthogonal frequency division multiplexing (OFDM). However, while EONs continue to provide large-capacity transmission for network users, they also pose new challenges to the flexible management and control of optical networks. Therefore, the combination of SDN and EONs (SD-EONs, the architecture is shown in the figure below) will be one of the powerful solutions for the optical control plane in the future. This article will briefly introduce the two aspects of fault recovery and resource allocation in software-defined optical networks.

Figure (a) SD-EONs network architecture; (b) BV-WSS structure; (c) OF-AG structure; (d) ER structure; (e) optical flow table

1. Fault recovery

Any network should have the ability to tolerate failures, and optical networks are no exception. Resilience is also one of the most desired attributes in optical networks. In SD-EONs, resilience can usually be divided into protection strategies and recovery strategies. Protection is a priori strategy, while recovery is a reactive strategy. Protection strategies are pre-planned and always exist regardless of whether network failures occur. Reference 1 provides a protection strategy for a ring network, which uses the priority of flow tables to pre-deliver high-priority working flow tables and low-priority protection flow tables. When the network is working normally, the high-priority flow table is used. Once a network failure occurs, it is immediately switched to the low-priority flow table to restore network communication. The advantage of the protection strategy is that the fault recovery time is very short because the entire process does not require additional information exchange. The disadvantage is that communication cannot be restored when the protection path fails again. In addition, it is very difficult to design a complete protection algorithm for large-scale and complex network topologies. The recovery strategy needs to design a method for sensing faults, and also an algorithm for fault recovery. Reference 2 uses detection packets to dynamically monitor the status of network links. Once a node or link fails, the optical proxy module will feed back the fault information to the controller. The fault recovery application in the controller runs the dynamic routing algorithm DAPSP (Dynamic All Pairs Shortest Paths) to restore communication by rerouting. Of course, in software-defined optical networks, controller failures cannot be ignored, because once a controller fails, it is likely to cause the entire network to be paralyzed. Reference 3 designs a master-slave controller solution to enhance the robustness of the control plane. The master and slave controllers periodically synchronize network status information. Under normal working conditions, the master controller is used to control the network. Once the master controller fails, the slave controller immediately takes over the control and management of the network.

2. Resource Allocation

EONs use Optical Orthogonal Frequency Division Multiplexing (OOFDM) technology to dynamically allocate spectrum resources on demand according to service requests of different bandwidths, and divide spectrum resources into many fine-grained frequency slots. At the same time, frequency slots can be split and aggregated according to the needs of the request, so that spectrum resources can be saved, thereby improving the utilization rate of spectrum resources. The most obvious feature of EONs is elasticity and variability. Transponders of multiple modulation modes can autonomously select the best modulation level according to the type of service. In SD-EONs, the design of routing and spectrum allocation algorithms should pay attention to the three constraints of spectrum continuity, spectrum consistency and spectrum conflict. Spectrum continuity: means that the spectrum slots allocated to each service must be continuous without gaps in the middle; spectrum consistency: means that for a service, the spectrum resources with the same serial number should be used on the optical link through which the optical path passes; spectrum conflict: means that a certain protection bandwidth is required between different services to ensure that the spectrum slots occupied by each service connection request do not overlap, and ensure that the signals do not interfere with each other during transmission and processing. Reference 4 designs an algorithm SEC-RMLSA based on Spectral Efficiency and Connectivity (SEC) to improve the service carrying capacity of SD-EONs.

Optical flow table

In optical devices, data forwarding should be operated according to the optical flow table. A simple optical flow table consists of input port (In Port), output port (Out Port), central frequency (Central Frequency, CF), spectrum slot width (Slot Width, SW) and modulation format (Modulation Format, MF). Here we introduce a simple method to extend the optical flow table based on Flow_Mod message.

Tips:

!: big endian storage; C: char, character type; B: one byte length, unsigned character type; I: 4 bytes length, int type; H: two bytes length; Q: eight bytes length; x: padding; 3x: 3 bytes of padding; 5s: 5 bytes of string. In optical equipment, data forwarding should be operated according to the optical flow table. A simple optical flow table consists of input port (In Port), output port (Out Port), central frequency (Central Frequency, CF), spectrum slot width (Slot Width, SW) and modulation format (Modulation Format, MF). Here we introduce a simple method to extend the optical flow table based on Flow_Mod message.

Step 1

Directory: ryu/ryu/ofproto/ofproto_v1_3.py

First, you need to redefine the protocol format in ofp_action_output, that is, redefine the original 6x padding field at the end to a format of 3 HHHs, which are used to store the center frequency, spectrum slot width, and modulation format respectively.

Step 2

Directory: ryu/ryu/ofproto/ofproto_v1_3_parser.py

Next, add three variables: center frequency, spectrum slot width, and modulation format to the parsing and serialization functions respectively.

Step 3

Catalog: In the application you develop.

***, load the three parameters output by the proposed algorithm to the end of the action. And send the encapsulated Flow_Mod message to the optical proxy. The result is shown in the figure, which is the Flow_Mod message captured by Wireshark.