LAS VEGAS -- For some companies, moving to the cloud can become a business imperative, particularly as industry changes fuel tectonic shifts in business processes.
Watertown, Mass-based Harvard Business Publishing (HBP), for example, found the ground moving beneath its feet. Its flagship publication Harvard Business Review, a monthly magazine that provides strategic and how-to information to business executives, needed to adjust to the era of digital media.
Other aspects of HBP's business also needed to scale, given the advent of trends such as distance learning and social media, explained Ken Griffin, director of IT services and operations for HBP, during his session "Preparing Your Application for Its Journey to the Cloud: Lessons Learned from the Application Battlefield," at Interop here this week.
"We made a conscious decision to reinvent our business. But the impact of that decision did nothing less than burst IT at the seams," Griffin said. "It blew us out of the water."
The changes prompted new technology and new websites. It also required the conversion of thousands of articles with tags for online search. As a result, HBP had to develop a "completely new digital organization," Griffin said.
Server hardware upgrade ends with cloud migration
While cloud computing made sense given HBP's goals, it was bad timing for a cloud migration project. It came on the heels of a major and costly server hardware upgrade -- to the tune of $3 million.
"Our hardware was in imminent death," said Griffin, who following the upgrade, vowed to make changes to the company's IT architecture to prevent a similar situation down the road.
The question was, 'Why move to the cloud to fix something not broken?'
director of IT services and operations for HBP
After the tumultuous upgrade, things had finally stabilized -- only to be upended by the plan to move to a hybrid cloud with Amazon Web Services (AWS) and VMware Inc. virtualization software.
"For the first time, we were in a good place. We solved our efficiency problem, and the plan was to take stock," Griffin said. "The question was, 'Why move to the cloud to fix something not broken?'"
New scalability and growth requirements prompted Griffin to make the move to cloud. For example, HBP needed to scale servers from 15 to 250 in a mere month. Investing in more hardware was an option, but the cloud became more compelling.
"We would move our applications to the cloud before it was time to upgrade our hardware again," he said. "We wanted to provide better services in three years' time."
Cloud migration: The demise of legacy apps
HBP first had to assess its applications. The portfolio included strategic and legacy apps. To make the transition to cloud, some apps would require code changes, but the ultimate goal was to put legacy applications out to pasture because they lacked strategic value over the long term.
"We categorized them as legacy apps, or 'technical debt,'" Griffin said. "The only thing you should do with this technical debt is to prioritize its demise. We put them on a path to extinction -- even some that generate revenue presently."
But as HBP laid the groundwork to migrate to the cloud, it simultaneously had to train and transition staff.
Not only did the project double the workload of the IT department, but it also required them to develop new skills -- fast.
"People, not the technology, was our biggest problem," Griffin said.
The IT department was basically going to have to double its workload, maintain the current environment, build out the new infrastructure, learn the skills required to do all of that, migrate all the applications and then decommission the old environment -- just after they had gotten through a bad time with a hardware upgrade, he said.
"Not only that; we wanted them to get excited about it," Griffin said.
Some Interop attendees questioned AWS issues and security, which Griffin addressed.
"The cloud is acceptable risk for us," he said. "We are looking at a small footprint outside AWS as well. At the end of the day, we're going to have another vendor we can access."
Three years into its five-year plan, HBP has its development apps in the cloud. In the final year, they will fully transfer everything to the cloud, and no services will be deployed in the traditional environment, Griffin said.
"We are not 100% in the cloud yet. But as one of my engineers said to me, 'We're almost kind of there -- and we will be,'" Griffin said.
No cost savings in the cloud
While the cloud migration is on track, the upfront costs are high.
Today, the project costs HBP double what it expects to spend long-term: approximately $30,000 for SunGard hosting services and $25,000 for AWS, added to its relatively recent hardware upgrade, including to Cisco Systems' FlexPod. But Griffin estimates that in two or three years project cost will be reduced to half of what it is today.
While HBP's migration story doesn't mirror every problem that Interop attendees experience in their own industry, it illustrates some common challenges other companies may face with a cloud migration.
"It's good to hear that others are having the same kinds of issues," said one software systems engineer with a large East Coast-based firm, who requested anonymity.
Like HBP, "We have a lot of data and a company mandate to get to 70% in the cloud by next year. But the truth is, some teams probably can; others can't. We probably won't make that deadline in every area."
Part of what attendees accept, though, is that the cloud brings inherent change: "I believe in two years' time, we will be in a better position to offer services to customers," he said.
Lauren Horwitz is an executive editor in the Data Center and Virtualization Media Group.