Q&A: Amazon Web Services' Simone Brunozzi

Simone Brunozzi (pic by Kaleel Zibe)

vnunet.com sat down with Simone Brunozzi, Amazon's web services evangelist, at the Thinking Digital event in Newcastle this week to discuss cloud computing strategies.

vnunet.com: Cloud computing is often talked about in the same breath as utility computing. Can you explain the two?
Simone Brunozzi: Utility computing can actually be compared to electricity. Originally it was produced in-house, but then people understood that, if you had a central power plant and could distribute it among factories, it was a good way to optimise and increase efficiency. If the power plant had a malfunctioning turbine, for example, you could shut one down while the others were still running. This big shift happened over 150 years ago. Now, with cloud computing, there is something similar happening. What was valid for electricity is now valid for computing; instead of having it in-house, you can use someone else's infrastructure in the cloud. You can have it on demand, you only pay for the functionality you need, and it's a very flexible resource, like electricity.

What are the key elements of a successful cloud computing model?
Whenever your business needs to innovate or create a new product, you need computing power. The biggest issue is scalability. Cloud computing should mean that, when you want it, you can have it in a short space of time. It means you can scale up when you need to, and down again when you don't need it anymore. When you add resources your performance should increase proportionately. You should be operationally efficient so that, when you add resources, the efficiency doesn't go away. And it should be resilient so that, if a single component fails, it is still available overall.

What are Amazon's 'seven lessons' in cloud computing?
We have learned in the past few years that the cloud needs to be designed for failure. It's what Amazon Web Services chief technology officer Werner Vogel said: if you design your system for failure, it will be able to react to any failure. The next thing that you need to think of is cloud computing as a loosely coupled system - a series of black boxes connected with each other, but in a way that a single box can scale up or down even if the others don't need to. It's also important to design for dynamism; a system using resources on demand will scale up, then down, then up again.

How about security in the cloud?
Security is everywhere - in the datacentre, at the network level - and you need to be aware of what tools there are to use them. There is always someone saying we can't do this, but you just have to think of a way to do it. If you need more RAM, for example, you can look at a shared database cache system. Or if you have database problems then maybe sharding [spreading databases across a number of distributed servers] will help, or database clustering. There is always a solution; you just need to find it. There is a huge cloud computing community so, whenever you don't know how to do something, just ask. I asked a question on Twitter about which tools to use for software development, and got an answer within 15 minutes.

Was it easier for Amazon to get a start in the cloud computing market because of the success of Amazon.com?
Amazon Web Services was created as a business because we had expertise in a few key elements - including large datacentres - and then we invested a lot on top of that. But we have a separate infrastructure from Amazon.com. We don't use the extra capacity of Amazon.com, as we have our own dedicated capacity; it's a low margin, high volume business. I don't know about other companies, but it makes sense for us.