The cloud has become integral to every business transformation strategy. Modern enterprise applications are fragmented, highly distributed, and becoming more and more cloud native. These applications are also more dispersed than ever before, across on-premise data centers, public clouds, and at edge computing locations in a truly multicloud setup.
Network and cloud teams are tasked with enabling these new type of business applications. As the world of network modernization evolves, network functions can run on the same cloud as IT and other workloads. This also means that network workloads need to be automatically provisioned to run wherever they are best suited. This evolution offers the promise of greatly reduced costs of operation as well as a new level of flexibility and agility, allowing customers to innovate and deploy new services as quickly and cost effectively as their competitors.
However, networks are becoming increasingly dynamic and many applications require low latency, with no time for human involvement available. Automation is the only way to efficiently manage services and applications across network and multicloud domains. (ok, so the title was disingenuous but now that you are here, keep reading) Automation is necessary in order to be competitive in terms of agility and cost.
By automating the service and application lifecycle management, customers can completely transform their business. Combining artificial intelligence (AI) and machine learning (ML) with automation, CSPs benefit from efficient and dynamic networks in a way that isn’t possible with human actions. Automation impacts the whole service lifecycle across:
- Design - Design creation and initial validation of network services based on application and business requirements
- Deploy - Intent-based orchestration of network services and applications
- Manage - Through data and analytics, closed-loop automation monitors the network
A network cloud operations model is driven by a software-defined representation of services and resources at various layers and the focus is on DevOps, continuous integration and automation throughout the lifecycle. The intent engine provides tools to declaratively model workloads and services. Then combine these building blocks into what is known as assemblies. This declarative based modeling approach allows the intent engine at deployment time to choose the best opinionated pattern from a wide range of instantiation options e.g., location, late binding resource choices, feasibility etc. The intent engine has logic that can decompose an assembly, configure all the dependencies needed to make it work, and orchestrate the instantiation.
Also, it instantiates the assurance components at the same time, crucially making the models available to assurance. This implies that there is a single process to develop, instantiate, scale, heal, terminate etc. from the simplest network function to the most complex service assembly. Allowing re-use of building blocks reduces both complexity and cost.
This approach enables the real-time feedback loop of communication between assurance and orchestration to enable zero touch operations. As depicted in the diagram there can be a series of feedback loops back into the orchestration layer.
- Incident detection and resolution
- AI-driven resolution of identified errors, with further auto diagnostics for unknown errors
- Sense and Respond to issues or opportunities for optimization
Assurance and AI is monitoring everything, across the full stack which is important as an infrastructure event can cause a series of errors. We are constantly looking to reconcile the actual and target state of the network and its workloads. Assurance and AI also help us to select an appropriate opinionated pattern that addresses the unique cause of the issues, rather than trying to heal all of its symptoms. We can leverage anomaly detection techniques to get ahead of a problem, and automatically trigger an unplanned change before it becomes customer affecting.
When designing networks for the cloud, a cloud native approach that encourages re-use is really important. Implementing an end-to-end network automation strategy is critical to managing next-generation networks. IBM Cloud Pak® for Network Automation, through its declarative modelling-based approach and closed loop operations, is helping enable that automation at scale.
Learn how IBM can help your business realize 82% reduction in costs to onboard new services and 6X reduction in customer service response time.