Big Data in the cloud

Abstract:

Big-data analytics has been, historically, an expensive proposition requiring millions of dollars of upfront investment. Since the point of big-data projects is to discover new things, their goals are often underspecified. This combination of big budget and lack of clear goals is a recipe for disaster (see Forbes’ article on “Why Big Data Projects Fail”). There are two trends that are making big-data projects less expensive, more agile and, hence, less risky. The first is the advent of open-source tools like Hadoop, Hive, Spark, R, Presto that cost a fraction of proprietary big-data software. The second is the maturing of public clouds that can run these big-data tools and change the economics on the hardware side and offer extreme flexibility. With these two developments, even small startups are able to run complex big-data projects and improve their products and services. In this talk, I’ll talk about what it means to run your big-data project in the cloud and talk about my experience building and running a big-data platform in the cloud.

Snacks and drinks will be sponsored by Orbitz.

Bio:

Sivaramakrishnan (Siva) Narayanan is currently working on his own startup. Previously, at Qubole, he made Hadoop, Hive and Presto play well in the cloud and saw the platform grow to processing 300PB a month. Earlier, he worked on query optimization and workload management in the Greenplum Parallel Database. He has a Bachelor of Engineering from BITS, Pilani, India and a Ph.D. in Computer Science from The Ohio State University. Siva has authored several patents and papers in the area of large-scale data management.