Designing for Disaster: How do you architect your systems to handle regional outages?

hey everyone, we've been thinking a lot about resilience lately, specifically how to handle major regional cloud provider outages. while we do have multi-region deployments for critical services, the complexity and cost are pretty high. how are you guys approaching architectural decisions to make sure your systems can handle a full region going offline. are you focusing on active-active across regions, or relying more on active-passive with fast failover. what strategies have you found effective for data replication and consistency in these situations without breaking the bank. i'm curious to hear what's working for you all.