We are reminded again today (Feb 28, 2017) about the fragility of the Internet. S3 is still down in us-east-1 and currently affecting pretty much every known business running on AWS and effectively shutting down the internet. I’m quite astonished at how a single service outage can bring popular other services like Slack, Dockerhub and even AWS’s own console to it’s knees. TechCrunch pointed out “Amazingly, even the status indicators on the AWS service status page rely on S3 for storage of its health marker graphics" There are known best practices like deploying into multiple regions to prevent these types of situations and the public cloud was supposed to usher in a new era of computing that allows for redundancy “built in”. However getting to multi-region isn't free and it takes foresight and work to make multi-region support.
Fortunately Spinnaker makes the process of implementing multi-region a lot easier. Spinnaker accomplishes by giving you a multi-region view of your applications which differs from AWS’s console which has a different view of each region. With Spinnaker, getting a global view of your application deployment is the default. Deploying into multi-region is as easy specifying what your ASGs (known as Server Groups in Spinnaker) will look like in those regions. While this doesn't remove the need to think about your stateful services like databases, but giving a global view of your service deployments make it a lot easier.