Intent-based networking enhances the operationalisation of network services by translating high-level business intents into dynamic network configurations.
Traditional networks enable services for customers through imperative commands. The service’s lifecycle follows a typical provisioning order workflow, where orders create, modify, or delete services. Each order is created and modified based on the necessary rework required for aspects of service provisioning parameters.
Intent-based networking provides services for customers through declarative commands that achieve the desired state of the network, supporting the intended use by customers. The service lifecycle is now supplanted by the lifecycle of the intent, which will last as long as the goals for the customer’s intended use of the network remain relevant to their objectives for the consuming service.
Expectations of an intent-based network
Intent-based network design augments automated networks to adapt rapidly to changing demands. Such a design assumes that some level of automation is already achieved in the networks and that the network demonstrates a level of elasticity to scale and accommodate fluctuating demands.
An intent-based network ensures that it consistently meets the desired service levels and quality of service (QoS) parameters defined by the intent. The QoS here would indicate how close the observed network state is to the desired state as defined by the intent.
Expectations of customers of intent-based services
As network functions shift to cloud-based services, more dynamic and flexible network provisioning must interwork with changing customer service demands. Customers demand faster and customised services with zero wait time to consume them. Additionally, the security of the network and services is maintained with every change occurring dynamically in the network.
Characteristics of an intent-based network
Intent-based networking uses Dynamic Network Resource Allocation (DNRA) to leverage such automation and AI to optimise network resource consumption based on high-level business intents.
At a high level, the following bounded contexts are required to design and implement an intent-based network, either as a layered architecture or a microservices / service-based architecture.
Analytics and Monitoring
- Provides real-time insights into network performance.
- Uses telemetry data to inform decision-making.
AI and Machine Lerning
- Predicts network demands and optimises resource allocation.
- Learns from historical data to improve future allocations.
Intent Engine and related Management and Orchestration
- Interprets high-level intents and translates them into network policies.
- Continuously monitors and updates policies based on feedback and analytics.
Automation
- Automates configuration changes and resource adjustments.
- Continuously monitors and updates policies based on feedback and analytics.
Network state reflects the configured behaviour for intent-based networking.
This topic is an in-depth topic for further updates in future posts.
Monitoring and security are maintained at all layers for intent-based networking.
This topic is an in-depth topic for further updates in future posts.
Use of Digital Twins
Digital twins are increasingly used in intent-based networking (IBN) to enhance network management and optimisation. More on this topic later
Defining and using Domain Specific Languages (DSLs) for Intent-based Networking
Domain-Specific Language (DSL) for IBN are specified to express intents in a human-readable and machine-executable way.
Key features include
- High-Level Abstractions
- Declarative Syntax
- Policy Definition
- Description of Network Topology and relationship between network elements (as in protected, diverse, etc)
- Declarations for validating intents against the current network state and policies
- Declarations for closing feedback loops that monitor network state and adjust configurations as needed.
Challenges with designing and maintaining Intent-based Networks.
Intent-based networking is not without its challenges.
Complexity with closed loops at multiple layers – Business, Service, Technology
Achieving closed-loop automation, a key goal of AN, relies on the network’s ability to translate intents into configuration actions, monitor the outcomes of those actions, and make necessary adjustments to ensure intent fulfilment. Such closed-loop automation requires advanced monitoring, analytics, and AI/ML capabilities at multiple layers to enable the network to learn and adapt to dynamic conditions.
Hidden side effects of closed loops – hidden commands and hidden states
Developing robust mechanisms for expressing intents unambiguously and enabling the network to interpret those intents accurately is essential for network stability and interoperability.
Advances in natural language processing, standardisation of intent models, and potentially domain-specific ontologies would benefit such defined expressions of intent and ensure shared understanding between users and the network.
Security and trust in intent-based systems
As network operations evolve towards greater autonomy, ensuring the security and trustworthiness of intent-based interactions should also grow in parallel. Security in intent-based networks includes protecting intent expressions from unauthorised modification, verifying the authenticity of intent sources, and implementing protections to prevent malicious or unintended consequences from automated actions.