Software-defined WAN is the most fundamental change in networking since MPLS took over from frame relay and Asynchronous Transfer Mode. While IT teams are able to manage their network via GUI-driven portals with cloud networking and security orchestration, SD-WAN is at the beginning of the technology stack journey.
AI and automation represent the next phase of capitalizing on extensive insights and data already generated by SD-WAN systems. For example, cloud AI is now used with SD-WAN intrusion protection to reduce false positives.
As SD-WAN technology evolves to provide more granular data and network insights, cloud AI will evolve to better understand the trends and consequences of that data over time, using collective experiences across multiple businesses. In many ways, SD-WAN represents the first step of the real transition to next-generation networking.
Some features, such as real-time SD-WAN path selection, provide automation. But IT teams must still make decisions based on reporting and engineer networks based on longer month-to-month analysis. The next step in automation with AI would eliminate the need for machine learning to make decisions this way. Instead, the future touts the ability to automate these decisions with more precision and speed, based on second-by-second data collection and insights.
How AIOps with automation offers a leap forward
According to Gartner, automation will operate more than 70% of network tasks, such as adds, moves and changes. This means faster delivery with fewer outages and reduced WAN issues. AI for IT operations (AIOps) and DevOps will have the ability to make changes on a 24/7 basis.
AIOps platforms offer automated virtual assistants that can work consistently on a company’s behalf to improve UX. SD-WAN vendors apply DevOps to combine both development and operations into a single entity to develop software features faster. These features can be integrated with AI automation.
SD-WAN is also evolving to simplify how customer DevOps teams use vendor APIs to create features within the platform. Teams can use APIs to implement certain automation services that address a specific business need.
Is end-to-end automation possible across all elements?
In isolation, SD-WAN automation does offer some significant benefits and enhancements. But, for the market to truly benefit from AIOps and automation, all SD-WAN elements will need to talk with each other via API integration. This communication will enable the system to make automated changes to WAN edge devices and reflect changes in the configuration of cloud resources, such as Microsoft Azure, AWS and Google Cloud. If all elements are brought together into a single, holistic approach, AI decisions and automation can be applied to the whole system instead of isolating each component.
Service providers and vendors also benefit from unified automation, which enables deployment and adds customer value across the SD-WAN overlay and underlay to bring everything together.
Accuracy with automation
Automation will also reduce human error when making changes. Almost all MPLS deployments over the last 20 years have experienced performance issues due to bad quality of service (QoS) configuration. Many customers have even removed QoS to improve UX. With AI and automation, QoS can be implemented correctly from the beginning and evolve as AI understands performance attributes over time.
Machine learning offers an answer across all technologies
SD-WAN currently delivers basic automation, such as error correction and path selection. To ensure automation provides significant benefits, AI must be driven to focus on business intent outcomes.
The intended result is to ensure the network operates against set policies, regardless of what changes occur. IT teams can deploy new thinking and strategies to achieve better business outcomes and UX. If an intended configuration doesn’t work or creates network issues, machine learning can identify the problem and either automate the correction or notify the IT team.
As more data becomes available, machines can use patterns from this data to determine which actions should be taken. The algorithms are given different sets of information to continue refining their decisions based on new inputs, even if the data has never been seen before.
In networks without AI and automation, adding new applications and topology changes often requires lengthy planning. When organizations migrate to the cloud or embark on new requirements, they usually face design complexity. SD-WAN automation will progress to a point in which adding complexity is simpler with drag-and-drop network elements and automated orchestration and testing.
Reducing costs with SD-WAN automation
Reducing costs is possible with SD-WAN automation because it helps reduce the potential waste of time and resources, which benefits network and security teams. Automation also helps keep costs down by eliminating the need to hire expensive contractors and engineers. While employees would likely need at least a few months of training before they can efficiently perform their tasks, systems can perform these tasks automatically.
The average SD-WAN platform requires knowledge and support, even with the transition from command-line interface to portal-based configuration. In many cases, SD-WAN vendors sell their offerings via integrators and managed service providers because of the expertise required for deployment. As automated configuration and management become more feature-rich and capable, most SD-WAN options will be DIY or co-managed, as fully managed services won’t offer the value they do currently.
SD-WAN automates many on-premises and cloud network tasks typically done by humans. The future will enable an intelligent, self-learning network capable of making decisions, while balancing various workloads in real time.