November 29, 2012, 2:15 PM — Late last year Daniel Bozeman, a software engineer at wireless analytics firm Mosaik Solutions, wanted to build a private cloud. The company is a heavy user of Amazon Web Services public cloud resources, and Bozeman had a vision of creating a system that would allow him to seamlessly run workloads either in the company's own data center, or in Amazon's public cloud.
He looked into OpenStack, but when he was investigating in late 2011, the feature sets just weren't there -- at the time the open source cloud platform didn't have a block storage service similar to Amazon's Simple Storage Service (S3), which it has since added.
Bozeman also looked at Cloud.com, which is now Citrix's CloudStack. But he wasn't too impressed with that either, particularly the price. Bozeman knew he wanted to use open source software, so VMware's vSphere and vCloud Director were out of the question.
He landed with Eucalyptus.
At Amazon Web Services' first-ever user conference this week in Las Vegas, Eucalyptus is attempting to carve outs its niche, and officials with the company have a simple message: "We're the private cloud version of Amazon Web Services," says President Marten Mickos. If you're looking for a private cloud that looks and acts like Amazon's cloud, allowing for easy connection with the public cloud, Mickos says Eucalyptus is your best bet.
AMAZON NEWS: Amazon launches cloud database with analytics tools, lowers S3 pricing
AWS even says so too, to a point at least. While OpenStack, and especially CloudStack, tout API compatibility with AWS, Eucalyptus scored somewhat of a coup earlier this year when it announced a partnership to develop AWS compatibility into its cloud product with the blessing of Amazon itself. That was enough to pique Bozeman's interest, and since then he's gone full on-board with Eucalyptus.
Mosaik now splits workloads that analyze and track mobile data usage and coverage between Amazon's public cloud and its Eucalyptus private cloud, using Chef recipes to provision resources based on certain workloads.
