agsandrew - Fotolia
In a presentation at the 2014 TDWI Executive Summit in Boston, consultant and author David Linthicum discussed the impact of cloud computing on business intelligence and data analytics -- and the opportunities it presents to organizations. Describing big data analytics in particular as "the killer app for cloud computing," Linthicum offered tips and advice on designing tiered data architectures for cloud analytics uses.
According to Linthicum, who is a senior vice president at Boston-based consultancy Cloud Technology Partners Inc., a solid architecture is the key to unlocking the potential benefits of cloud platforms. Linthicum said many cloud-based BI and analytics systems lack a well-grounded architecture, which he argued isn't the fault of IT architects, but rather a result of the corporate focus on short-term -- and often shortsighted -- plans. "Most [businesses] are on a quarter-based system, where the ability to meet the numbers becomes paramount," Linthicum said. Instead of thinking tactically, he encouraged companies to work to establish cloud data architectures holistically by applying the same infrastructural underpinnings enterprise-wide and on the same timeline, rather than focusing efforts where it's most immediately convenient without considering the long term.
Another common mistake that Linthicum warned against is the tendency for businesses to simply throw new technology at their problems in hopes of fixing data architecture difficulties. Constructing a cloud analytics architecture is "just like building a house," he explained. "The big question is not, 'What kind of wood are we going to use to build the house?' It's, 'What's the house going to look like?'"
Also, moving from a complex and dysfunctional on-premises architecture to one in the cloud won't resolve the underlying problems or improve functionality. Linthicum urged businesses to review their data analytics architecture and consider rebuilding it if necessary before transitioning to cloud computing. Pushing things out into the cloud without a lot of thought "can be as dangerous as doing nothing," he warned.
With a well-planned strategy for cloud architecture, Linthicum said businesses can leverage the tiered approaches that cloud computing supports, potentially separating their BI and analytics data into public, private, community and hybrid clouds based on what fits best for particular applications. In addition, smart organizations "are typically picking three or four cloud providers," he said; that way, "if one goes down, the others can pick up the load." By doing so, businesses can spread their dependencies across different vendors and data models, which helps provide more flexibility, agility and architecture scalability. Linthicum also advocated implementing a service-oriented architecture (SOA) for analytics as part of cloud deployments.
Overall, he thinks that cloud computing can play a positive role in big data analytics applications and other BI and analytics processes.
"Before we were limited by the capabilities of technology," Linthicum said. "Now we're limited by imagination, because the technology is there." But, he added, a cloud analytics architecture must be properly planned to be sustainable -- and a business case needs to be made: "You don't shift to the cloud for the cloud's sake. There has to be a reason."
Watch a video report on the gap between Hadoop hype and adoption, as highlighted at TDWI Boston
Read more about what went on at the TDWI summit, with a look at the benefits and challenges of agile data warehousing
Listen to a podcast with TechTarget editors discussing the burgeoning relationship between data architecture and the cloud
Dig Deeper on Big data analytics