Archived Content

The following content is from an older version of this website, and may not display correctly.

Cloud evangelists often preach about server-provisioning speeds and high infrastructure-utilization rates enabled by cloud computing, but cloud can also tell a story of easier post-acquisition integration.

Following US pioneer ISP AOL’s acquisition of the famous left-leaning news site Huffington Post, announced in February, the infrastructure that supported the Post had to be integrated with AOL’s infrastructure.

The private-cloud environment the buyer recently built made the integration a relatively painless process, absorbing the news site’s environment in about one week, according to Mike Manos, AOL’s recently hired VP of technology operations.

Manos’s colleagues Jacob Rosenberg and Aaron Lake presented on AOL’s internal cloud at this week’s Uptime Institute Symposium, organized by the Uptime Institute in Santa Clara, California. Rosenberg manages AOL’s content-services group and Lake is the company’s director of infrastructure development.

To be sure, the speed-of-deployment story is there.

"When I first got to AOL 10 years ago, it could take 6-12 weeks to put a server on the floor," Rosenberg said. Today, the average time between the moment a user submits a request and the moment a VM instance is created is 10 seconds. The VM is usually up and running within a few minutes.

In addition to reducing provisioning time, AOL’s goals in building the cloud were to reduce expenditures associated with expansion of IT capacity, to expand capacity for existing and new applications, as well as to be able to scale applications dynamically.

Dynamic scalability is a key capability for AOL, many of whose web properties sit dormant, with relatively low traffic rates, until there is a promotion for one that causes traffic to go through the roof for the time of the promotion.

The company also launches a lot of what Rosenberg called "pop-up sites" – sites created specifically for certain events. The royal wedding in the UK, for example, was one of these events. "We’re not going to have another royal wedding probably for at least a couple months," he said jokingly.

Another example would be sites that cover electoral races in the US.

The goal to increase capacity was achieved as well. Since the infrastructure supports eight VMs per 1 physical server, AOL’s capacity increase was eightfold, Lake said.

AOL started the project after a major restructuring that took place in January 2010. Following the restructuring, the company realized that it had both staff and infrastructure constraints – a realization that led to the idea to create an internal Infrastructure-as-a-Service solution.

AOL owns and operates three data centers in the US, totaling about 200,000 sq ft of raised floor. Two of them are in Virginia and one is in Silicon Valley. The company also uses several colocation facilities outside of the country.