Dispelling 13 Common SDN Myths
Misunderstandings can occur in all aspects of life and cutting-edge technology is no exception. We explore 13 common myths about SDN with the aim of providing a better understanding of this critical technology.
Myth 1: OpenFlow is the only SDN standard
This is a common misunderstanding as OpenFlow was one of the first SDN standards defining the communications protocol in SDN environments, and it has already been successfully deployed by vendors such as Google on white-box [or un-branded] equipment. However, OpenFlow is not the only protocol available or in development for SDN. Other networking protocols have emerged such as PCEP, NETCONF, and BGP-LS, which can run an SDN environment, particularly in carrier scenarios. Besides OpenFlow, there are also mainstream open-source SDN controllers such as ONOS and ODL that support southbound application program interfaces (APIs).
Myth 2: Only white-boxes can be used for SDN
White-boxes are not unique in their support for SDN forwarding, traditional networking equipment can also be used. In addition, white-boxes wonít completely replace traditional networking equipment as there are complexities in running and managing white-boxes and it will be some time before they are widely adopted. For now, traditional networking equipment remains the most important type of forwarder for SDN.
Myth 3: SDN is a new type of NMS (network management system)
SDN has three core features: separation of the forwarding plane and control plane; centralized control; and open programmability. Itís important to distinguish between centralized control and centralized management. The latter can be achieved using traditional NMS and OSS. However, centralized control involves centralized scheduling of dynamic resources and route computation, thereby simplifying the network to gain visibility of the entire network. While SDN should incorporate traditional NMS FCAPS capabilities to form a closed loop operation (monitoring->control->configuration), it is not a new generation of NMS.
Myth 4: SDN is all about the controller
According to ONF, SDN architecture consists of: SDN applications, which are programs that communicate behaviors and required resources with the SDN controller via APIs; the SDN controller, which is a logical entity that receives instructions or requirements from the SDN application layer and relays them to the networking components; and SDN networking devices that control the forwarding and data processing capabilities for the network. Therefore, while the controller is a key application in SDN that manages flow control to enable intelligent networking, SDN is more than just the controller.
Myth 5: SDN will replace all traditional equipment
SDN does not apply to all networks and in some cases traditional networks are more suitable due to their reliability through distributed routing features. Additionally, SDN does not require replacement of all equipment within each layer on a specific network. Essentially SDN is about reconstructing network architecture from a distributed control network to a centralized control network and centralizing distributed control protocol that originally ran on separate equipment.
Myth 6: SDN only applies to data centers
In the past, the rapid development of SDN was driven by the automation of data center networks. However, SDN now has much wider applicability such as in carrier networks. SDN can help carriers simplify O&M complexity, improve network resource utilization, rapidly expand or change services to meet customer demands, and drive business growth opportunities such as cloud services and IoT.
Myth 7: SDN matters to cloud service providers
Although early implementers of SDN were mostly cloud service suppliers, enterprises and carriers are also gaining significant benefits from SDN. In addition to delivering high performance and minimizing cost, SDN is flexible, scalable, easy to manage and secure to support their businesses in the future. SDN is developing rapidly, drawing major players into a thriving SDN eco-system including equipment manufacturers, cloud vendors, telecom carriers, OTT vendors, ISVs and more.
Myth 8: SDN applies only to switches
In the early days of SDN, the technology was associated with switches because OpenFlow technology is applied to switches. However, the core SDN feature of separating the control plane from the forwarding plane also applies to routers, optical transmission equipment, and wireless products.
Myth 9: Software development engineers will control the network
As SDN brings software programmability to the network, traditional network planning and operational professionals may be concerned that SDN will reduce their responsibilities and weaken their control of the network. In reality, SDN will reshape network planning and operation and bring different levels of automation. This means the core duties of network technicians will evolve to include service planning, the creation of different policies and templates, and helping software engineers understand and develop network applications. While, SDN will bring changes to network operations, the network should still be controlled by network professionals.
Myth 10: SDN is only about automation
SDN has successfully enabled data centers to achieve automation, but it delivers more than automation. By centralizing the control plane, organizations can simplify network management by unifying network resources for optimized resource allocation to accelerate agility, reduce the cost of managing network resources, ensure availability of business critical applications, and improve the customer experience.
Myth 11: SDN is unsecure due to centralized control
SDN is by no means the first technology to introduce centralized control. There are many central control applications such as BGP RR already deployed on traditional networks. Furthermore, a distributed control and management network has more security attack points, whereas centralizing control consolidates scattered defenses to improve overall security. However itís important to note that a centralized controller will become an attackers' main target, therefore a sophisticated security design is essential for SDN deployments.
Myth 12: SDN requires a special SDN chip
Although a special "SDN chip" is often mentioned, SDN does not require any special chip, and there is no dedicated ASIC commercial chip for OpenFlow or SDN. It is possible that equipment manufacturers will customize chips to better serve SDN requirements, but such a chip is not mandatory. Equipment manufacturers can optimize a traditional chip to ensure that the equipment has a southbound API to comply with SDN requirements.
Myth 13: SDN is treated separately from NFV
The NFV reconstructs network elements to decouple network functions from network hardware. SDN reconstructs the network as a service to change the delivery method of network connections. Just as vehicles (NFV) need roads (SDN), SDN and NFV are inexorably related. Another popular concept is NFV reconstructs functions from layer 4 to layer 7, while the SDN reconstructs functions from layer 1 to layer 3, but they are two sides of the same coin.