Impala Lightning Talks in NYC (Special Strata + Hadoop World 2015 Location)

Great news -- we're taking the meetup on the road for Strata + Hadoop World New York 2015! Thanks to Collective (http://www.collective.com/main.html) for graciously agreeing to host.

After the customary libations we'll proceed with a trio of Lightning Talks:

Talk 1:

Using Impala as a Service Backend: What We’ve Learned (15 mins)

Chris Ingrassia, Senior Director Engineering, Collective

Impala clearly has value as a SQL-on-Hadoop tool for performing analysis and running ad-hoc queries over the data you already have in Hadoop, but what happens when you take the plunge and try to hook part of a service into it over JDBC and use it as you might a “traditional” database? In this presentation, we will review what Collective learned through the implementation of Collective’s internal reporting service, Vega, which makes extensive use of Impala in conjunction with Spark, Parquet, and Spray across a 165 node YARN+Impala cluster and roughly 13 billion rows of data.

Talk 2:

Support for Nested Types in Impala (15 mins)

Marcel Kornacker, Chief Architect for Data Technology, Cloudera & Alex Behm, Software Engineer, Cloudera

Impala 2.3 includes support for complex schemas, aka nested types (containing arrays and maps). This talk will give an overview of the extended SQL syntax and some preliminary performance results, comparing the flat relational TPC-H schema with its corresponding nested schema.

Talk 3:

Impala Resource Management with YARN (15 mins)

Matt Jacobs, Software Engineer, Cloudera

Impala 2.3/CDH 5.5 includes a number of new improvements that better enable Impala to share cluster resources using YARN. This talk will contain a brief overview of resource management options for Impala users today, improvements we've made in the CDH5.5 release, and how to use Impala with YARN successfully.