AWS offers object storage and you can integrate it with EMR. This requires import. They also offer another option for s3 storage which can be used instead of HDFS, but it cannot be accessed as the regular object storage. It is the same story with Azure - Azure Storage and Data Lake Store.
But reading Google's product page, it seems that you can run map/reduce and Spark on their regular storage. How come?
But reading Google's product page, it seems that you can run map/reduce and Spark on their regular storage. How come?