Intend based analytics (IBA)

Intend based analytics

A proper IBN system uses intent-based analytics (IBA) to stay aware of network changes in real time, continually ensuring that your services remain in compliance with your intent. That’s intent assurance.

Change management:

IBN system shall provide change management.

There are two types of changes normally we observed in the network.

• Uncontrolled changes: Changes that happened in network automatically without touching the network is called uncontrolled change. For example, connectivity breaks for any xyz reason or interface gets down certainly.

Three IBNS features are essential when dealing with uncontrolled changes:

• Built-in analytics that can detect deviations from expressed intent
• Probes that pull essential data and state from the network
• Fast root cause analysis
• Controlled changes:  In controlled changes, the day to day activity and execution of configuration in planned window etc. Basically we are aware of these changes in the network.

IBNS features that are essential when dealing with controlled changes:

• Reliable pre- and post-deployment verification
• Fast, accurate rollback capability
• Flexible network probes
• Fast root cause analysis

Discovering Actionable Insights: The key to making sense of all the data you gather is to put it all into real-time, queryable context.

• You must be able to detect conditions of interest. You need analytical probes that ask essential questions and root out important information, stripping away what’s irrelevant.
• Conditions of interest, in turn, are classified by the relationships between conditions.

Probing Your Data:

A good IBN system enables you to define your own probes.

An IBN system should provide a library of prebuilt probes that you can use as-is or modify to your needs.

Probe flow can be represented as below:

A simple example of a probe is

1. Collect information from X, Y, and Z.
2. Analyse that collected data to understand what’s normal over time.
3. Create a baseline from that normalized data.
4. If telemetry shows a drift off of the baseline of more than this defined threshold, alert me.

Root cause analysis:

Root Cause Identification (RCI) analyses the relationships between anomalies in your network to trace back to a root cause.

Configuration rollback: An IBN system must support the feature of configuration rollback feature based on date and time. We should be able to restore the service if something is broken due to control changes.

Key thing to keep in mind when considering IBN system

1. Don’t Start with Hardware
2. Free Yourself from Vendor Lock-In
3. Sort through the Fluff
4. Don’t Rely on Home-Grown Solutions
5. Automation Is Only Part of the Story
6. Rid Yourself of Calculated Guesswork
7. Empower New Ways of Thinking

Key benefits of IBN system

• Reduce time to delivery by 90%.
• Eliminate 85% of all outages by in-depth verification before deployment.
• Lower OPEX by up to 83%
• Lower capital expenses (CAPEX) by up to 60 %
• Lower mean time to resolution by 70%

IBN FAQ

SDN	IBNS
SDN is networking architecture.	IBN is used to manage network lifecycle. I.e. plan, design, implements and operate.
It’s abstraction model of physical network	IBN handles a network whether it is SDN based on non SDN based.
It takes generic configuration and pushes it to device specific configuration.	Whereas IBN takes input as intend and convert it into a device specific configuration.
Just like basic configuration automation system, SDN is part of IBN	A complete IBNS system consists of entire network lifecycle management components that includes plan, design, implement and operate. It also responsible for SLA compliance and auto remedies.

Orchestrator	IBNS
Orchestrator is used to orchestrate the entire IT component that may include compute, storage and networking.	IBN is used to manage network only.
Orchestrator is not only integrating with IBN systems, it also integrates with other controller of other domain.	An ideal IBN system shall be integrated with orchestrator as part of overall orchestration process.