I recently asked six industry experts their opinions on higher-performing technologies now available to make the most of big data. Read the perspective of Mark Troester, CIO & IT Global Product Marketing Manager at SAS, below and watch other views from Expedia, Teradata, Elder Research and Target Marketing in these video clips – view now in Flash, or on YouTube.
In this big data era, what are your thoughts on new technologies that combine higher performing systems with analytics?
MARK TROESTER: With more and more companies taking advantage of big data, many technologies have come about to support it. Hadoop is certainly all the rage, and there are others, such as in-memory, in-database and grid technologies, that can be applied to big data implementations. But the most important consideration is to build a flexible infrastructure that leverages the right mix of these technologies – it’s simply not one size fits all.
First of all, the architecture should be driven by the business goals and requirements. Its design should account for things like real- or right-time access, and support all relevant data types and multiple device types, including mobile, embedded analytics, etc. This generally leads to a mix of technologies and approaches that allows an organization to meet future needs.
Some analytic initiatives can benefit from factoring in all available data into the analytics modeling and deployment – in fact, not only all of the data needs to be factored, but also massive numbers of analytic variables to perform analytics at a very granular level. For example, performing pricing optimization at the individual product SKU and store level while accounting for all customer and product attributes will likely provide better results than performing pricing optimization at a broad product category or geographic region level based on minimal attributes. In other scenarios, it may make sense to use analytics and organizational knowledge to filter massive amounts of incoming data.
As you can see, there are many approaches to handling big data. Overall, the key is to wade through all of the hype associated with these hot technologies and build a flexible infrastructure that will provide robust analytics.