SAN FRANCISCO -- At GigaOm Structure 2011, the big theme was just getting businesses to consider adopting cloud computing. This year, it's clear that the cloud has been accepted (for the most part) as the future of computing.
See also: Akamai CEO: We're entering 'Instant Internet' phase Panel: Bring applications to big data, not vice versa Cisco CIO discusses understanding big data in context
Now the debate is about which cloud vendors and strategies you should choose, whether you build something in-house, go with a major vendor like Amazon Web Services, develop a mixture of the two, and much more. Honestly, the choices can be overwhelming for both small businesses and large businesses stuck on old infrastructures with tight IT budgets.
Executives from three of the biggest digital media properties at the moment -- Evernote, Dropbox, and Netflix -- offered advice to audience goers while commenting on their own cloud growing pains.
"You don't always know what you're going to build and whether or not its going to work," said Aditya Agarwal, vice president of engineering at Dropbox. "You need to build an infrastructure that lets you build a lot of things in parallel."
At a higher level, Agarwal advised that you want to set up really small teams and give them autonomy to build quickly.
Alexei Rodriguez, vice president of operations at Evernote, commented that the company doubled in size every six months last year, with the Asia/Pacific market playing a huge roll in that.
"Asia, from a growth perspective, has been pretty phenomenal for us," Rodriguez said, explaining that tackling China, in particular, is presenting a major challenge to scale up resources to handle that region.
Dropbox, too, has grown internationally considerably as of late. Agarwal acknowledged that Dropbox hasn't re-tuned its infrastructures for each market, which might be considered a mistake as he said that the cloud storage provider might change this strategy in the future.
Adrian Cockcroft, director of Netflix's cloud architecture unit, described Netflix's cloud strategy evolution -- a story that is likely relatable for companies of all sizes.
Of course, it's important to point how how big Netflix grew in the last year alone, from serving roughly 25 million users in North America to expanding to Latin America as well as in the United Kingdom and Ireland.
The point is that Netflix started off not knowing what would work for them, which is probably the case for most businesses. Cockcroft said that Netflix engineers realized that it would need multipledata centers to do that kind of global availability, but it didn't feel like it had the necessary experts in-house to do that.
"At the time, we basically started to try risks and try things out," Cockcroft said. Questions arose about how to move data into the cloud and access it as well as how to deploy these systems. Explaining most of Netflix's current cloud architecture, Cockcroft cited that the online video giant stores data in Cassandra, with continuous archives on Amazon S3 as well as another copy to a non-Amazon-hosted site.
Cockcroft commented that it was a very similar story to Facebook because the cloud team also wanted to maintain a focus on developer productivity and autonomy.
Interestingly, while all of the panelists were very in favor of sampling with different providers and infrastructures, everyone also expressed support for Amazon Web Services.
"For companies both big and small, the point of comparison shouldn't be old infrastructures but something like AWS," Agarwal asserted, arguing that if you're building the infrastructure in-house, it should be easier to use than AWS. If it's not, then you have a problem.
Rodriguez added a caveat that you should at least be very clear about why you would choose one method over another. Nevertheless, he also posited that companies joining the cloud should also likely start on AWS and build out from there.