How we build a fully regional cloud architecture on AWS
At least for me, building on the cloud gives often gives me unexpected joy. To think that I can deploy servers in Bahrain or Japan from my couch is still something I get excited about!
At Apptrail, we run completely on the cloud, specifically AWS, and one of the main reasons is how easy it is to launch services in new geographical regions. We deploy our services to independent cloud regions and provide our customers with regional endpoints (e.g. events.us-west-2.apptrail.com).
Cloud providers have a concept of regions for their cloud. These are isolated geographical regions where their physical infrastructure is hosted and where they offer cloud services. For example, at the time of this article, AWS offers 26 different cloud regions. Software applications built on the cloud can leverage cloud regions to also build a concept of regions in their applications to improve their services.
To understand regionality, it's good to understand cellular architecture. Cellular (or cell-based) architecture is a way of separating systems into isolated cells to reduce the blast radius from something going wrong. The idea is that a failure shouldn't ever cross cells, and so cellularization is a way of creating those independent partitions. The actual method of choosing a cell can be anything, from random based on ID to logical like geographical region. You can also have nested cells for further isolation (e.g. Large volume customers in US West region could be a particular cell). AWS Availability zones, for example, are also sub-regional cells. All in all, cellularization is a very powerful way of designing available systems, you can learn more about it here.
