One Big Data Strategy, Three Parallel Processing Platforms
Let’s face it: big data is still very new. And that means the infrastructure platforms on which big data analytics are performed are also relatively immature.
What it doesn’t mean, however, is that organizations can’t or shouldn’t pursue big data success. They simply need to be resourceful and calculated as they step into the big data waters and evolve their strategies and platforms over time.
Intel IT, for example, previously employed a “one size fts all” approach to business intelligence (BI) and analytics using a centralized enterprise data warehouse (EDW). However, with data volumes growing—much of it unstructured data from proxy, machine, server, and access logs—and more business groups wanting to mine value from those datasets, there was an emergent need for new analytical capabilities and less expensive platforms.
"Due to its cost and the type of data it contains, our EDW is largely reserved for structured enterprise data that has horizontal use across the entire company,” says Chandhu Yalla, BI engineering manager for Intel IT. “To be honest, we don’t want to pollute the EDW with unstructured data and vertically-focused, one-off analyses.”
Intel IT’s overarching BI and big data goal remains constant: provide the right data to the right people at the right time. The methods for achieving this goal are evolving to accommodate an increasingly wide variety of business use cases, analytics, and data types.