Cloud computing infrastructures will routinely fail, panel says

Most of us can agree that we're still in the nascent stages of cloud computing, and that there are a lot of pitfalls to overcome still. But many Silicon Valley players argue that we need to accept the failures of the cloud to learn from them.

The major theme being discussed at GigaOm's Structure conference in San Francisco this week is the future of cloud computing. Most of us can agree that we're still in the nascent stages of this technology, and that there are a lot of pitfalls to overcome still. But many Silicon Valley players argue that we need to accept the failures of the cloud to learn from them.

While moderating a "guru panel" dubbed "What can the Enterprise Learn from Webscale?" on Wednesday, Facebook's VP of Technical Operations Jonathan Heiliger said that "web businesses are designed to fail," meaning that the fundamental components are going to fail, but that the apps should be built to under scale this failure.

Claus Moldt, Global CIO and SVP Service Delivery at Salesforce.com, added to that:

Everything within the infrastructure needs to be designed with failure in mind...That's how you have to run your business.

However, the panelists agreed that the best way for a business (especially startups) to avoid complete failure when transitioning to the cloud is building on a horizontal scale - not a vertical one.

LinkedIn's VP of Engineering Kevin Scott VP warned that if companies don't think about the infrastructure on a scalable basis in the beginning, they'll "deeply regret it."

In addition to having fundamental fault tolerance, you have to think very quickly about how you're going to scale your applications.

It's not just about economics...it's about good system design, dealing with networks and components failing...It forces you to expose clean APIs between these components.

Sid Anand, a software architect at Netflix, offered the example that when the video streaming site moved to Amazon's Web Services cloud, noting that the move was successful because developers rebuilt the apps from scratch with a new DNA. Thus, as Anand mentioned, Netflix had an "infinite scalable and resilient" system because they didn't "shoe-horn" apps into a new platform that they weren't originally designed for.

Of course, one of the biggest hindrances to the cloud still for most consumer and enterprise computer users is questionable security. Without delving too much into specifics, Jacob Rosenberg, an architect at Comcast, advised that enterprises need to take a more holistic approach to understanding security. Scott added that this could incorporate "traditional security techniques and aggressive monitoring" by IT departments.

Noting that "there is a new generation of measures that are being taken to secure cloud-type services," Rosenberg said that these ideas can still be applied to existing hosted products to make them more secure as well.

Related:

Newsletters

You have been successfully signed up. To sign up for more newsletters or to manage your account, visit the Newsletter Subscription Center.
Subscription failed.
See All
See All