Cloud computing technology is maturing, as is the practice of storing and analyzing data there. While privacy and accessibility concerns still remain, they no longer completely eclipse features like elasticity and quick deployment.
"There's a lot of momentum for the cloud," said Tony Cosentino, vice president and research director for Ventana Research, an IT research and consulting firm based in San Ramon, Calif. "I'd expect to see a continuation of this trend in BI [business intelligence] and analytics."
The keyword is continuation. Success stories about analytics in the cloud are becoming easier to find, but barriers still exist, Cosentino said. And for one analytics business, overcoming those barriers required a shift in its culture and pricing model.
Movin' on up in cloud technology
Razorfish Inc., a marketing and technology company headquartered in Seattle, didn't really choose the cloud; instead, the cloud was kind of thrust upon it. Founded in 1995, the startup was eventually bought and sold by companies that enabled it to collect and analyze online marketing data from customers behind a secure firewall.
That was until Publicis Groupe S.A., based in Paris, France, purchased the company in 2009. With that acquisition came a new way of doing things, such as processing data using Amazon Web Services (AWS). After that, certain types of cloud analytics also migrated to the cloud.
Read more on cloud computing
Are cloud computing cost savings worth the wait?
Read up on the top 10 public cloud services businesses want to buy
Find out how cloud shops are misled with high SaaS data integration costs
"We took proprietary processes and built those using Amazon Elastic MapReduce," said Matt Comstock, vice president of BI for Razorfish.
Elastic MapReduce, which uses a hosted Hadoop framework within the Amazon Elastic Compute Cloud (EC2) infrastructure, gives Razorfish the ability to scale up or down as needed. The feature comes in handy when processing large volumes of clickstream data, which tends to increase dramatically around the holidays because of a peak in shopping traffic, according to a published case study by AWS.
"As clients come and go, we needed to have an infrastructure where we didn't have to put a huge amount of capital up front to get the capability we needed to process the data," Comstock said.
Razorfish uses Hadoop and MapReduce for preprocessing data, but other tools, such as Teradata Aster for data warehousing and Tableau for data discovery, come into play at different points in the analytics process. Today, all of that work takes place within AWS, Comstock said, comparing it to a service-oriented architecture.
Snap back to reality
The cloud works well for the company now, but getting there required Razorfish to deal with its fair share of headaches.
For one, Razorfish needed to explain the change in infrastructure to its clients. In some cases, the news prompted customers to request to meet with Amazon and discuss its privacy policies and standards. But, for the most part, selling that news turned out to be pretty straightforward, Comstock said, though he did note Razorfish tends to work with marketing departments and not IT.
"They don't want to know how the sausage is made," Comstock said. "They just want the sausage. And if you can get it to them faster, great."
In an ironic twist, the harder sell turned out to be the price tag. Part of Razorfish's initial pitch was to deliver the same service to its clients more reliably and for less money. It was a good concept in theory, but it looked very different in practice.
"It's not as cheap as they sold it," Comstock said.
After migrating to AWS and paying for the service monthly, the company soon realized its pricing model needed to be adjusted to incorporate the true cost of the technology, which can be hard to make tangible when it's tucked away as a capital investment.
"Typically, we can't figure out how to take a capital investment and distribute that on a per-client basis and figure out what to charge," Comstock said. "Now we can look at it from a labor and infrastructure perspective to figure out a fair allocation."
The realization was made after the company migrated to the cloud, which provided an unfortunate egg-on-the-face moment. In midstream, the company had to circle back to its clients and explain why its prices would be increasing. In contrast to the Amazon ads announcing how cheap the cloud is, Razorfish's message made for a tough -- albeit necessary -- sell, Comstock said.
"It's a good thing," he said. "It's just [a] painful transition period."
Cloud computing also called for a cultural shift. Little things could add up quickly, and the company had to start thinking differently about how it was using the technology. "It's as simple as developers doing code and leaving the machines on all night," he said.
Still, for all of the rough patches, having technology that seamlessly scales with the business has been advantageous, according to Comstock. And, although he calls outages a concern, he also has faith in Amazon's IT department.
"You're getting a lot of good practices … that would take you time and money to get done internally," he said. "That's definitely a benefit."
Nicole Laskowski is the news director for SearchBusinessAnalytics.com. Follow her on Twitter: @TT_Nicole.
This was first published in November 2012