Clients of Amazon Web Services experienced a second major outage due to electrical storms in the last few days.  The power was restored to the data centre environment about nine minutes later but Netflix, Instagram, Pinterest, Heroku and other companies that depend on its infrastructure were affected.

I agree with Mike Lee from ZD Net that the outage issue was less to do with Amazon, and more to do with individual companies not using cloud services to their full potential, to create geographically redundant links.

Lee quoted Intelligent Business Research Services advisor Jorn Bettin: “They could operate at a higher level of redundancy, so that these sorts of outages would only have a minimal impact on them. It’s a matter of cost.”

This therefore raises the topic of managing levels of redundancy. It is possible to solution workloads to have 100% availability in the cloud, but this will be at a cost.

The reality is that many organisations, particularly small businesses, will look at their bottom line, and decide that the cost to mitigate the risk isn’t worth maintaining 100%.  They are therefore making a choice to balance risk with the price of 100% uptime.  Even though the cloud service provider may not provide 100% uptime, in reality, given that the provider’s data centre technology is more advanced with high levels of redundancy than most clients’ own data centres, the provider has higher availability.

Over time the reliability and availability of cloud services will improve as data centre technology evolves. The risk of cloud service downtime will exist for some time, so businesses should make a conscious choice on the level of availability of their workloads and select a cloud service provider that will manage to this.

Cloud providers need to be upfront with clients on what availability they provide. In New Zealand, this disclosure is assisted by the Cloud Code of Practice.  This may mean selecting multiple cloud service providers or selecting a provider with geographically diverse infrastructure, with failover capability.

I’d be keen to hear your thoughts on my suggestions and about the challenges and opportunities of cloud computing in general.