If you are in the Australian IT industry it will not have escaped your notice that AWS has had an outage in a single availability zone in Sydney. This has created a large amount of coverage online, some outrage, some opinion and a lot of WTF moments.
The volume of coverage goes to show how successful public clouds have become.
I cannot comment on the infrastructure of any public cloud let alone AWS. Before I worked at Microsoft I certainly used AWS and am familiar with it. But I cannot and would not like to comment on how it is built and managed, the truth is like the majority I do not know and to be honest I don’t care. Amazon is a very large, very successful company and therefore it’s easy to assume they do a lot very right indeed.
I’m sure they have fully redundant systems, like all public clouds, and probably more redundant systems than you or I could shake a stick at. However nothing is 100% and the availability of any system is the aggregate of the provided uptime of each component. If that aggregate does not meet your business requirements you need to architect in redundancy to meet the accepted risk level the business is prepared to tolerate.
How many of these environments were too reliant upon a single data centre? Too many by the looks of things.
When natural disasters hit they do tend to impact a city hard. Look at Sydney today, Brisbane a few years ago.
Your cloud deployments need to be different. You must architect for cross region redundancy or make the business aware of the risk. Let the business requirements drive the strategy, cloud, hybrid or multi vendor. After all it’s what we are here for to drive successful business outcomes.
Disclaimer: Please note although I work for Microsoft the information provided here does not represent an official Microsoft position and is provided as is.