This week, I have been inundated with articles on the terror that was the AWS outage. I have read everything from an article touting the downfall of AWS to promises that their solution could have avoided the disaster all together. Well, as with most things the reality is neither that cut and dry or dire. The truth is even with the best planning, the greatest environments can go down. Like it or not, we live in a physical universe and disaster recovery is still a mandatory requirement for those organizations with mission critical data.
Let me begin by first extolling the virtues of virtualization and cloud. It is changing the world we live in by allowing people to leverage more resources at a cost that was not thought possible even five years ago. We use it every day in almost every way. From the cash register at a store to the most complex of database applications, just about everything we do is driven by virtualization and data.
Now the bad. Because of that reliance we also feel the impact of an outage in a way that we never have before. A mistake in a simple line of code could literally take a third of the internet down and with it make many people’s data unreachable.
So what then is the answer as more and more companies move to the cloud? If we want to enjoy cost savings, are we doomed to suffer from an unstable and risky environment? Or must we continue to overpay in the hopes of achieving a greater level of stability? Could it be that as we achieve efficiency through consolidation that we lose resiliency and peace of mind? In the new cloud world order is an outage now just a total outage?
Well just like the cloud has been around for decades in different iterations (remember grid computing?) the answer to true resiliency remains the same answer as always. Whether it is called COOP, or DRCP, or just good old redundancy, the reality of cloud is that while its name denotes something unfathomable in the sky, it is actually a series of servers living in a physical data center. As such it is tied into and made available to networks not inclusive to its own LAN. And you know what? There is no reason you can’t tie two clouds together just like you network a computer.
For all the propeller heads out there, this is just good old contingency planning. And for everyone operating within a cloud-always world mindset, this simply means to leverage multiple providers. This gives you greater diversity in bandwidth providers, elasticity, data portability, and in the worst of scenarios a true offsite that will allow you to operate without interruption or at least minimal downtime.
The reality of cloud is that it is a truly great way to grow your business and maintain a high level of efficiency as well as to maintain a good cost basis. The reality is also that it is not a silver bullet and as I have written in the past, there is still no substitute for good planning. So before you pull everything out of your cloud provider or put everything into a single location, pull together a true plan for redundancy. The cost is usually not factors higher, but sleeping at night without worry and making sure your business continues to run could be priceless.