Amazon Web Services’ scalable infrastructure for lean IT

Planet Lean: How does Amazon Web Services enables companies to use lean IT?

Carlos Condé: In a nutshell, by giving our customers core IT components they can use to build any kind of system, from storage services to databases, without having to worry about developing the infrastructure itself. Additionally, our pricing model is pay-as-you-go, which means you can play around with building blocks, build something, and see if it works for you.

PL: Upon developing this solution, how did you ensure you were responding to a real customer need?

CC: Amazon Web Services started with Amazon S3 in 2006. The feedback we got from that helped us to come up with Amazon SE2, a service that offers virtual machines running on the cloud. People started building systems using that. They could now build databases, so we realized we could also offer a database service. Essentially, by listening to users and looking at what they do with your services you can develop more products – we now have over 35.

PL: During your talk at the Lean IT Summit, you encouraged the audience to “embrace failure” – what did you mean by that?

CC: In the past, failure in IT was very expensive, because hardware was. Amazon Web Services is based on pay-as-you-go, which means that in the face of an unsuccessful project you will only have lost, say, a couple hundred bucks. You haven’t invested money to get started, which enables you to embrace failure, so to speak. Startups can suddenly concentrate on business value, but also on their growth.

PL: What does low-cost innovation look like?

CC: When you have low cost building blocks, you can try whatever you want. Take Amazon Redshift, our data warehouse service. Previously, only large organizations could afford similar services. Now, small companies can too. Sometimes you don’t know what ROI you are going to get from deploying a certain solution, so being able to give one a try without having to take many risks is critical. To increase innovation, you need to lower the cost of failure.

PL: Amazon is now able to deploy new code every 11.6 seconds – what sort of infrastructure is needed to do that?

CC: It comes from a necessity: Amazon scaled to millions of users around the world, and needed more and more machines to do that. The key was automating. An advantage of running on the cloud is that you can treat infrastructure as code, automating the deploymenet of software on machines exactly how you push new versions of your software.