T-SDN: the road to the future of Optical Transmission

In the next few years, carriers will use SDN technology to enable their man and wan to carry cloud services. Therefore, they need a multi-layer SDN architecture to optimize the routing network and transmission capabilities, this increases the flexibility, scalability, and efficiency of man and wan.

Preface

Cloud Computing technology has completely changed the design and operation of data centers (iDCs. SDN technology introduces this on-demand concept to the network: allows applications to consume network transmission resources just as they use virtual computing and storage capabilities. In an SDN-based data center network, network connections are no longer permanent, and they are stimulated by virtual machines (VC). After virtual machines are disassembled, they no longer exist. When cloud computing technology extends across man and wan to multiple data centers, new applications based on the distributed cloud architecture also expect operators to use SDN technology to upgrade their networks, allow them to consume network transmission resources as needed.

Figure 1 impact of cloud architecture and cloud services on Transmission Networks

It is not easy to apply SDN Technology in man and wide area networks. Different from the simple network topology of the data center, a single vendor structure, and unlimited bandwidth resources, the operator's network topology is generally more complex and the suppliers are more diversified, more restrictions such as bandwidth and technology. Any attempt to accelerate and automate the transmission between man and Wan must be consistent with the existing service deployment system. The dynamic consumption mode of distributed cloud services will put huge pressure on the existing network operation modes based on static and predictable traffic. Operators need to adjust the current traffic engineering sequence through dynamic resource management to ensure that the transmission network can improve network efficiency and provide business protection and high applicability in the face of rapidly changing cloud service connection requirements.

The entire communication industry has begun to evolve towards SDN. The corresponding technical suggestions include network abstraction and open private data and transmission layer capabilities to external control through the OpenFlow protocol. Although this is a good starting point, carriers need a more scalable SDN architecture to achieve greater value, this architecture needs to be able to span multiple layers (including the optical layer, electrical layer, Ethernet layer, IP layer and ip vpn service layer) deploy and associate topology, transmission resources, and physical constraints. Only a multi-layer SDN architecture can have full-network views and unified control capabilities, achieving the SDN ultimate vision: increasing the flexibility, scalability, and efficiency of man and wan.

Reasons for deploying SDN in man and wide area networks

Traditional man and Wan are designed for relatively static and predictable network services: enterprise branches that connect to a distance and enterprise headquarters in the center responsible for business processes. Network Service deployment is generally implemented through complex IT/OSS systems using APIs of different vendors at the lower layer, and manual configuration is performed through the command line interface or script file. Carriers are willing to spend days or even weeks doing this because they believe that these network services will survive for a long time. Now, the emergence of distributed cloud architecture will inevitably have a huge impact on this network operation model.

For example, once the basic transmission network facilities are put into use, enterprise users require that the transmission services of the man and the network be dynamically established and deleted in seconds like data centers. They require carriers to ensure fast and on-demand delivery of transmission services that meet latency, jitter, and bandwidth requirements. For example, when virtual machines (VMS) dynamically move based on the availability of virtual computing and storage capabilities, enterprise users require operators to provide different network connection services for daytime Network Service Processing and night data backup. At this time, the carrier can consider providing bandwidth or booking bandwidth as needed.

Carriers need a fast and cost-effective way to provide advanced network services for enterprise customers. Therefore, they need to abandon this time-consuming manual service deployment model, in addition, it is necessary to quickly provide new network services to improve the competitiveness of enterprises. To meet these requirements, carriers need standardized high-level APIs to abstract networks.

When the network traffic model becomes more dynamic, operators begin to face the complexity of the introduced network operations, and the originally several months of network planning cycle will no longer be applied to ensure service quality. To ensure that network services do not use up bandwidth or are out of Preset QoS, operators need to review network resource allocation and conduct traffic engineering decisions more frequently.

When you map network service requirements to transmission resources, the operator needs to perform real-time network analysis to meet the customer's SLA requirements and ensure that the transmission resources are used in the most economical way. To assist in Real-Time Route Selection and computing, operators need to be able to provide real-time full-network views and deploy and transmit resources as needed, however, these multi-layer full-network views and control capabilities cannot be implemented for various reasons today.

Transmission Networks optimized for cloud services

To carry cloud services, carriers need a multi-layer SDN architecture to divide the network into two main components: the network virtualization automation component and the programmable transmission network.

Figure 2 Alcatel-Lucent network SDN architecture optimized for cloud services

Alcatel-Lucent believes that the main function of the programmable transmission network component is to provide high-reliability and high-performance transmission resources (spanning the optical layer, electrical layer, Ethernet layer, IP layer and IP service layer ):

-Dedicated NPUs and optical transmission devices provide the most economical sorting, switching, and forwarding capabilities from Layer 1 to Layer 3

-Industry-proven distributed control protocols are embedded into network hardware to Ensure network scalability, stability, and protection.

-Open southbound API interfaces (such as OpenFlow, NetConf, SNMP, RADIUS, and DIAMETER) allow applications based on standard interfaces and SDN controllers to monitor and control network resources of different vendors at any level.

-New network capabilities, such as segment routing, can support new cloud services through offline route computing

Alcatel-Lucent believes that the network virtualization and automation components provide high-level abstract network views for applications, and have real-time full-network view and control capabilities. This component can be scaled to the maximum extent at the lowest cost on a general CPU platform:

-An SDN controller is used to configure network elements on one or more networks.

-Policy-driven business deployment allows network tasks and services to be defined as policies, so that network services can be deployed more quickly and on a larger scale.

-Policies can integrate many low-layer network tasks into high-level network functions to hide the transmission complexity of different vendors. Through Network abstraction, operators can achieve faster innovation and better competition

-Resource management and control include resource manager, ALTO, and PCE.

Application Cases

Like many new technologies, SDN technology is introduced out of economic considerations. Before making any investment decisions, carriers need to verify that SDN can bring new business increases, improve efficiency and flexibility, and reduce CAPEX and OPEX. As described in the following application cases, SDN can accelerate cloud service deployment and support new business models without affecting the operation of existing network services.

On-demand data center interconnection Service

Problem: the customer needs their network and bandwidth services to dynamically adjust according to the movement of the data center virtual machine. While complying with existing network and security policies, they need to dynamically support changes in network traffic patterns, such as data backup at night and data bursts.

Solution: allows customers to dynamically manage network connection services between data centers. This new business model allows customers to re-allocate the purchased bandwidth between data centers, or provide on-demand extra bandwidth based on changes in the network traffic pattern. This new business model also allows customers to define specific business requirements, such as bandwidth, time, and network connection time.

SDN role and role: This new service can be quickly defined through a policy-based abstract framework. Customer requirements can be mapped to business policies and then stimulated by some form of events, such as when the virtual machine responsible for backup is created. Then, the policy-based business deployment framework starts to automatically call network resources, such as increasing the size of the optical business ODU particles and disabling the transmission service when the demand disappears. The Resource Manager manages bandwidth to ensure that the transmission service is not over-used.

Create an on-demand business chain

Problem: End-to-End cloud services, including IT and network connections, are often distributed across multiple data centers, this makes it very difficult to integrate end-to-end cloud services for remote users. Because the routing needs to be manually changed, such business implementation takes several weeks.

Solution: Accelerate advanced business deployment and reduce operation costs by dynamically creating business chains for each business combination.

SDN role and role: PCE can be used to calculate the optimal route in the middle of a virtual machine to integrate network connections, thus canceling time-consuming manual route editing. In this process, you can use network policies to map each user to the man wide area network service chain optimized for them. The network topology of the service chain is stored in the SDN resource manager instead of the transmission device. This avoids network hardware upgrades because the transmission device requires a large amount of CPU resources.

Dynamic Route Selection and traffic targeting

Problem: network operators and cloud service providers must consider intermediate transmission networks when deciding how to define traffic and how to deploy IT resources. This information can be used for a group of specific users to decide which video cache resource to use on the network and which data center the virtual machine should use, you can also find out which public video cache or data center can provide additional resources when local resources are exhausted.

Solution: allow users to declare the connection requirements of the two connection points, such as latency, jitter, and bandwidth. Then we provide a way to measure the path and business that meet such network attributes in real time.

The role and role of SDN: multi-layer network analysis allows ALTO play the biggest role. For example, from the perspective of optical transmission, a direct optical channel between the two connection points is also the best choice for the city. However, at the IP level, the path through an intermediate node may be better, because it can provide more bandwidth when the latency needs to be met. Peer congestion and cost data can be used to determine which peer is most suitable for connecting to a public data center or video cache. These mappings can be automatically deployed at the transmission layer, or delivered to the content delivery network (CDN) or Cloud collaboration layer (Cloud Orchestration ).

Dynamic Multi-layer network optimization

Problem: As network traffic becomes more dynamic and unpredictable today, the time window for network traffic engineering is too scattered, often for a few months, leading to a decline in the availability and quality of network services. Even if a small network change occurs, the operator must perform isolated monitoring on each network layer and analyze the implementation results after Network Association online.

Solution: provides a policy-based Approach to track SLA and automate traffic engineering at all layers of the network. The connection, attributes, and bandwidth of the transmission service can be dynamically adjusted in minutes or months.

SDN roles and functions: third-party applications can continuously access and analyze the network-wide topology and status information stored in the resource manager. The analysis results are used to dynamically trigger network policies to redeploy network resources, such as adjusting the ODU container size or ing between services and optical channels. In this way, operators can avoid resource exhaustion, improve user experience, and optimize bandwidth usage.

Traffic Engineering-based business plane

Problem: carriers need a simpler and more effective way to allocate network transmission resources in different types of business.

Solution: allows operators to create and manage multi-layer business planes based on traffic engineering in the form of slices, and ensures that these business planes meet the requirements of transmission businesses. For example, data backup of an enterprise or data center can be mapped to a large bandwidth business plane that is not in use. You can disable the optical layer FEC in a service plane to increase bandwidth and reduce latency. In another service plane, you can open the optical layer FEC to provide services for businesses that are not sensitive to latency and packet loss.

SDN roles and functions:

The Service plane can be created manually, and the application creation can be optimized through the network mentioned above. You can create a network policy to map the network traffic of different users to the business plane.

Conclusion

The emergence of cloud computing and cloud services is changing the network construction mode of operators. SDN was initially used only for virtualization and automation of the data center network. Now it is also used to optimize cloud bearer in man and wide area networks. Although some work has begun to abstract the transmission network through OpenFlow, this is not a complete SDN solution. Only like Alcatel-Lucent, the SDN architecture across multiple layers (L0-L3) can have full network view and unified control capabilities, in order to increase the flexibility, scalability and efficiency of man and Wan in the cloud era.