What is network orchestration?

Network orchestration is the automation of networks across different types of infrastructure devices, network domains, and even multi-vendor systems with a network. In addition, orchestration can be policy-based or event-driven automation through the use of programmatic interfaces like RESTful APIs that are enabled by third-party software or a network infrastructure solution using a single end-to-end integrated solution.

As expected there is some confusion and lack of clarity around the differences between network automation and network orchestration.

What is the difference between network automation and network orchestration?

In simpler terms, network orchestration is an advanced level or form of network automation. Let us think of network automation as a manager in charge of low-level single “if this then that” tasks, whereas network orchestration controls high-level sequences of tasks that are interdependent on each other across various systems.

Network automation is often straightforward i.e. linear., it carries out tasks based on device states and configurations. In Network automation , Programming is done through command lines or third-party script-driven frameworks while in Network orchestration programming is carried out through programmatic RESTful interfaces like API calls.

Now we’ve successfully differentiated between Network automation and Network Orchestration. 

Let’s go a little deeper into the different types of orchestration.

Types of Network Orchestration

We have three main types of orchestration, these include Policy-Based Automation, Software-Defined Networking, and Intent-Based Networking Systems. 

We’ll explain each and its differences in this article.

‍

Policy-Based Automation (PBA)

Policy-based automation does not take into consideration individual device parameters but instead, it controls device “policies.” The Policies are then applied to each specific device type or role. This is arguably the simplest type of orchestration that can lead the way to “software-defined” orchestration.

Policies can also be duplicated between devices or supporting systems, and such duplicated policies can be used to create a template that can be implemented at new facilities as long as the schema remains untampered. For example, We can often find policy-based automation in GUI dashboards since it’s generally considered one of the easiest ways to implement changes across multiple devices.

Software-Defined Networking (SDN)

Software-defined networking includes a software-based controller or APIs. These controllers permit programmatic management of functions, which includes provisioning, monitoring, and configuration. This tends to be more high level than policy-based automation but not as high as the intent-based networking systems.

Compared to policy-based automation, software-defined networking can be programmed directly, also the application and controller layers can be separated, creating room for increased flexibility and changes to the network in real-time.

Network intelligence is controlled in a single centralized location where devices check-in for the latest updates, instructions, and configurations. Considering the flexibility of virtualization and software a lot of the SDN can be vendor agnostic, simplifying design and creating a more complex and fluid ecosystem.

Intent-Based Networking Systems (IBNS)

AI is used in intent-based networking systems to optimize the network for a human-specific intent. Instead of coding, network administrators define intent as a result or business goal. Machine learning (ML) is used by artificial intelligence (AI) to optimize the network for that intent by establishing routines, setting policies, and responding to system events.

IBNS is regarded as the highest level form of orchestration because it performs much of the technical work on its own and allows humans to fine-tune the ML algorithms to improve their performance. Much of the manual work in IBNS is done during the ML algorithm training and intent translation.

The IBNS is given instructions, which the algorithm interprets as the network’s overall goals. This could be specific service level agreement (SLA) thresholds for device groups and applications, as with Celona’s unique MicroSlicing technology, or specific QoS rules that the organization wishes to always maintain.   

Following approval, the AI deploys the changes it believes will correspond to the intent of the instructions and establishes a feedback loop to audit the network over time. Data will be made available to both the ML algorithm and administrators over time in order to determine how effective the intent-based network is at any given set of instructions.

While this might arguably be one of the most effective ways to deploy orchestration at scale, it is entirely based on the speed with which machine learning algorithms can achieve new milestones, which can take considerable time depending on the intent.

‍

Network orchestration use cases

Network orchestration offers many benefits, including:

1. Ensuring QoS policies are being followed
2. Automating troubleshooting
3. Auto-configuration of new devices
4. Establishing overlays for the control plane and the forwarding plane
5. Provisioning network services
6. Providing workflow automation
7. Orchestration doesn’t just empower enterprise Wi-Fi networks, but private LTE and 5G networks too.