Over the past few years, cloud-native applications have gained significant traction within organizations. These applications are built to work best in a cloud environment using microservices architecture principles. Everest Group research suggests that 59 percent of enterprises have already adopted cloud-native concepts in their production set-up. However, most enterprises continue to operate traditional networks that are slow and stressed due to data proliferation and an increase in cloud-based technologies. Like traditional datacenters, these traditional networks limit the benefits of cloud native applications.
SDN is the network corollary to cloud
The network corollary to a cloud architecture is a Software Defined Network (SDN.) An SDN architecture allows decoupling of the network control plane from the forwarding plane to enable policy-based, centralized management of the network. In simpler terms, an SDN architecture enables an enterprise to abstract away from the physical hardware and control the entire network through software.
Current SDN performance is sub-optimal
Most of the current SDN adoption is an afterthought, offering limited benefits similar to the lift-and-shift of applications to the cloud. Challenges with the current SDN adoptions include:
- Limited interoperability – Given the high legacy presence, enterprises find themselves in a hybrid legacy SDN infrastructure, which is very difficult to manage.
- Limited scalability – As the environment is not designed to be SDN native, applications end up placing a high volume of networking requests on the SDN controller, limiting data flow.
- Latency issues – Separate control and data planes can introduce latency in the network, especially in very large networks. Organizations need to carry out significant testing activities before any SDN implementation.
- Security issues – The ad hoc nature of current SDN adoption means that the entire network is more vulnerable to security breaches due to the creation of multiple network segments.
SDN native is not about applications but about the infrastructure
Unlike cloud native, which is more about how applications are architected, being SDN native is about architecting the enterprise network infrastructure to optimize the performance of modern applications running on it. While sporadic adoption of SDN might also deliver certain advantages, an SDN-native deployment requires organizations to overhaul their legacy infrastructure and adopt the SDN-native principles outlined below.
Principles of an SDN-native infrastructure
- Ubiquity – An SDN-native infrastructure needs to ensure that there is a single network across the enterprise that can connect any resource anywhere. The network should be accessible from multiple edge locations supporting physical, cloud, and mobile resources.
- Intelligence – An SDN-native infrastructure needs to leverage AI and ML technologies to act as a smart software overlay that can monitor the underlay networks and select the optimum path for each data packet.
- Speed – To reduce time-to-market for new applications, an SDN-native infrastructure should have the capability to innovate and make new infrastructure capabilities instantly available. Benefits should be universally spread across regions, not limited to specific locations.
- Multi-tenancy – An SDN-native infrastructure should not be limited by the underlay network providers. Applications should be able to run at the same performance levels regardless of any changes in the underlay network.
Recommendations on how you can become an SDN-native enterprise
Similar to the concept of cloud native, the benefits of an SDN-native infrastructure cannot be gained by porting software and somehow trying to integrate with the cloud. You need to build a network native architecture with all principles ingrained in its DNA from the very beginning. However, most enterprises already carry the burden of legacy networks and cannot overhaul their networks in a day.
So, we recommend the following approach:
- Start small but think big – Most enterprises start their network transformation journeys by adopting SD-WAN in pockets. This approach is fine to begin with, but to become SDN native, you need to plan ahead with the eventual aim of making everything software-defined in the minimum possible time.
- Time your transformation right – Your network is a tricky IT infrastructure component to disturb when everything is working well. However, every three to four years, you need to refresh the network’s hardware components. You should plan to use this time to adopt as much SDN as possible while ensuring that you follow SDN-native principles.
- Leverage next-gen technologies – To follow the principles of SDN-native infrastructure, you need to make use of multiple next-generation technologies. For example, edge computing is essential to ensure ubiquity, AI/ML for intelligence, NetOps tools for speed, and management platforms for multi-tenancy.
- Focus on business outcomes – The eventual objective of an SDN-native infrastructure is better business outcomes; this objective should not get lost in the technology upgrades. The SDN-native infrastructure should become an enabler of cloud-native application implementation within your enterprise to drive business benefits.
What has your experience been with adoption of an SDN-native infrastructure? Please share your thoughts with me at [email protected].
For more on our thinking on SDN, see our report, Network Transformation and Managed Services PEAK Matrix™ Assessment 2020: Transform your Network or Lie on the Legacy Deathbed