HA != DR

HA = High Availability.
  The ability for your application (or infrastructure) to survive the failure of one or more components and continue to operate.

DR = Disaster Recovery
  The ability to recover from a failed state.

I was recently in a discussion where the idea came up that if you have HA in the cloud you don't need DR.  I used this opportunity to remind everyone that even in this new world of on-demand, scalable, redundant infrastructure the age old mantra of "always have a backup" still holds true.

The discussion centered around AWS and if you have redundant services running your application, spread across multiple regions, the catastrophic loss of an entire region wouldn't affect the running application.

The above statement is true, to be highly available your application should be multi-component, multi-availability zone, multi-region, multi-provider even.  But the question brought the discussion to an interesting place.  If you have that level of HA, why would you need DR?

A few years back a company learned that lesson the hard way.  In 2014 a hosted source code repository company called CodeSpaces, billed as "full redundancy", had all of their data permanently deleted by a hacker and went out of business.  How?  The hacker got login to their root console and pressed the delete button.  Not to diminish the need for proper security and the need to protect your root account, it was not having a Disaster Recovery plan which was the fatal mistake.

If after reading about that hard lesson learned and still not convinced that something like that could happen to your product because of all of the redundancy and resiliency which is built into your system, you should think of DR as an insurance policy.  You have it on your house, your car, your life?  Why not put an insurance policy on your job?  Accidents do happen.