freshidea - Fotolia

Get started Bring yourself up to speed with our introductory content.

AWS and OpenStack hybrid cloud a balancing act

Using a mix of services from OpenStack and AWS, one company gained efficiency and agility -- without sacrificing control.

Mention public cloud to IT teams and many immediately think of cost savings. And while it's true, moving certain applications to AWS, Google or Azure could save companies money, the almighty dollar is not always the driving factor -- nor does cost drive decisions on which provider to choose.

Lithium Technologies, a social software provider in San Francisco, began migrating customer-facing applications to the public cloud late last year. They considered a number of cloud providers, including Google, Azure and Rackspace, but ultimately went with the provider they were most familiar with, said Justin Franks, lead cloud engineer at the company.

"Some of our engineers were very comfortable with AWS, and it made sense to use it as a first step," Franks said.

While that familiarity made it "easier to adopt AWS" for Lithium's foray into public cloud, adopting it required a bit of work behind the scenes. Lithium's developers had to work around a few application resiliency problems to ensure the apps could handle a cloud environment -- where instances go up and down. The teams performed resiliency testing to ensure that services met the service-level agreement Lithium provides to its customers and all associated security requirements.

New applications were easy to migrate to AWS: Developers built them from the ground up to take advantage of the cloud infrastructure. Legacy applications will need to be rewritten from a different environment to work in a public cloud, Franks noted. "We have to rewrite those applications to be a little more resilient and work with technologies like Consul and Chef. Those [applications] will require a little more effort," he said.

To get the public cloud project off the ground, developers weren't very concerned about cost optimization, Franks noted. Now that apps are up and running, IT staff are going back and cost-optimizing across AWS using in-house and third-party tools.

"The [cost monitoring] tools in AWS weren't sufficient enough because they didn't provide enough in-depth analysis of the virtual machines we were using," Franks said.

Lithium was also concerned about vendor lock-in, so Franks is cautious of using several technologies from a single vendor. IT staff there believe they should be able to move applications to most cloud providers within a reasonable amount of time.

"If AWS has a problem, we may need to turn up applications with cloud provider B," Franks said.

Finding that sweet spot for hybrid cloud

Despite those few hiccups with application resiliency and cost optimization, Lithium benefited from efficiency in its public cloud move -- cutting development times down from weeks to minutes. And Franks anticipates significant capital savings on infrastructure.

"But that's a double-edged sword when you're dealing with economies of scale," he cautioned.

"As we start to ramp up the number of servers in AWS, that cost per unit may start to exceed the capital and operating expenses of running your own hardware in a data center and putting our own cloud on top of that," Franks said. Building a private cloud signals "an intersection where you start saving money," he added.

Openstack is not something you want to tinker with, especially if you're going to use it in production.
Justin FranksLithium Technologies

That's when Lithium built its hybrid cloud -- a mixture of AWS and OpenStack private cloud, plus a number of other public cloud providers. Once the company passes that "intersection," it will rely less on AWS and expects to gain significant savings running things on OpenStack.

Lithium uses OpenStack in production to carry the workload of applications and services.

"We're looking at it to fulfill the role of our traditional data center, but with a layer of virtualization and agility on top of a traditional data center that's not VMware," Franks said.

But putting OpenStack into production wasn't as easy as it was to move or build applications in AWS. And Lithium developers didn't have the familiarity with OpenStack they had with AWS. So, it sought help.

"As with any new technology, you want to wait so it's not bleeding-edge. Yet you want to use it soon enough to gain competitive advantage," Franks said, adding that Lithium played it "by the book."

The company did some research and development in OpenStack with its own resources to determine if it made sense to go it alone or use a third party to help move workloads out of R&D and into production. The company opted for option B -- engage OpenStack experts and third parties to help.

"OpenStack is not something you want to tinker with, especially if you're going to use it in production. There are a lot of pieces to it," Franks said. Lithium developers are also in constant contact with other companies using OpenStack in development, which, he added, was "a very small percent."

As for hiring OpenStack experts in the future? Franks wouldn't comment directly, but said the company is always looking for talent.

About the author:
Michelle Boisvert is Executive Site Editor for SearchCloudComputing.com. Contact her at mboisvert@techtarget.com.

Next Steps

Creating a successful cloud project

OpenStack adoption hindered by service woes

Building an enterprise private cloud with OpenStack

OpenStack takes private cloud market

A glimpse into the future of AWS

This was last published in January 2015

Dig Deeper on Public cloud providers

PRO+

Content

Find more PRO+ content and other member only offers, here.

Join the conversation

1 comment

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

One of the more dicey aspects of Hybrid cloud is the relative performance of the platforms. I have seen research from 451 (powered by Krystallize Technologies) that suggests machines provisioned with the exact same technical specifications can produce highly variable performance. This implies Justin Franks and others must be judicious when moving workloads among platforms. For example, development may take place in-house, but moved to AWS for production. While the target machine can have the exact same specifications, the actual platform layer support may be as little as a third of the capacity needed for the same level of performance. Without the visibility into the cloud or hosting platform, you will not see this until it is too late. Moreover, conventional metrics will fail to highlight the real issue.
Cancel

-ADS BY GOOGLE

SearchServerVirtualization

SearchVMware

SearchVirtualDesktop

SearchAWS

SearchDataCenter

SearchWindowsServer

SearchCRM

Close