We can learn a lot from this about distributed, decentralized systems and their resilience. Caching as much and as local as possible, while accepting that the processes might have no network connectivity is key. This way each node, and whoever can reach this one node, can work independently for long stretches of time.
