Google launches Cloud Dataflow, says MapReduce tired

Google on Wednesday launched Cloud Dataflow, a big data analytics service to crunch information in either streaming or batch mode.

The announcement, made at Google's I/O keynote in San Francisco, helps round out the search giant's cloud stack, which is adding features as it aims to take on Amazon Web Services.

Urs Hölzle, senior vice president at Google, outlined Dataflow and a demo revolved around crunching Twitter data and sentiment around World Cup games. Dataflow was the headliner in a series of cloud services outlined. 

Hölzle said that Dataflow has replaced MapReduce inside Google. Cloud Dataflow is designed to analyze pipelines with "arbitrarily large datasets."

"Cloudflow does for entire pipelines what MapReduce did for single flows," he said.

Roughly speaking, Google's Cloudflow would line up against Amazon Web Services Redshift, a datawarehouse service, and AWS' Elastic MapReduce, which uses Hadoop to crunch large datasets.