Week 4 Posting - Subnetting in the Cloud
Something interesting when looking at subnetting in the cloud is the amount of translations that need to happen in the backend for everything to work. With the numerous customers that a cloud service provider like AWS would be hosting, there would need to be the ability to have completely isolated, but still routable RFC1918 networks sitting beside all their other customers. When looking through the AWS documentation there doesn't seem to be a limitation on what IP addresses can and can't be used, and as I have spun up my own networks it is very flexible. Multiple layers of routing need to happen for this all to work. The customer routing, the datacenter infrastructure routing, AWS internal routing, and their connections to the outside world. Since AWS allows customers to provision their own networks, this has to happen seamlessly in the background using templated provisioning. It would be a wonder to see what this architecture looks like and how AWS can keep up with it.