Skip to content
- People think about ML when they think about predictability. This is only one dimension
- Intelligent Cloud Networks have taken a lot of analysis and backed into their fabric. So you don’t see it but it is running
- Sometimes it is trained by humans, sometimes it is auto trained
- Automation is another form of prediction because you are removing humans from the equation
- Static automation is not going to work or is not the answer, with automation you need an intelligent controller that can take the network changes into account and behave differently next time. For that, a feedback loop is important that can be done by an intelligent controller or control plane. This new knowledge can then be pushed to data-plane elements
- Some predictions are written on the wall. For example the need for a backbone. So in some cases, we can say that these intelligent cloud networks are removing the need for lot of unknowns so the need for predictability is decreased.
- Coming back to a network backbone, any network engineer can easily predict that enterprises need a cloud network backbone. You don’t need any ML to tell you that. So build it. This is called human predictability. Avoiding the known issues.
- For example, I can almost predict when my son when he comes back from school and needs food immediately, so I would prepare it in advance. Similarly, network humans are also known and can predict what is needed from a cloud network. We need to use that knowledge. We can easily predict that multicloud network is a must so why not build our networks with that in mind to begin with.
- In future these intelligent cloud networks platforms have the potential to bolt to a neural network and use ML to train an AI engine.
- There is no simple answer or pointing toward a product here. It is the collective wisdom. What I can say that networks have become more efficient. They have quickly adopted to new world. The complexity has gone up. APplications are more complex. Closed systems are open system.
Comments are closed