Page History

...

This is an HDFS root directory under which Hive's REPL DUMP command will operate, creating dumps to replicate along to other warehouses. 

Blobstore (i.e. Amazon S3)

A set Starting in release 2.2.0, a set of configurations were was added to enable read/write performance improvements when working with tables stored on blobstore systems, such as Amazon S3.

hive.blobstore.supported.schemes

• Default value: s3,s3a,s3n
• Added In: Hive 2.2.0 with HIVE-14270

List of supported blobstore schemes that Hive uses to apply special read/write performance improvements.

hive.blobstore.optimizations.enabled

• Default value: true
• Added In: Hive 2.2.0 with HIVE-15121

This parameter is a global variable that enables a number of optimizations when running on blobstores.
Some of the optimizations, such as hive.blobstore.use.blobstore.as.scratchdir, won't be used if this variable is set to false.

hive.blobstore.use.blobstore.as.scratchdir

• Default value: false
• Added In: Hive 2.2.0 with HIVE-14270

Set this to true to enable the use of scratch directories directly on blob storage systems (it may cause performance penalties).

hive.exec.input.listing.max.threads

• Default value: 0 (disabled)
• Added In: Hive 2.2.0 with HIVE-15881

Set this to a maximum number of threads that Hive will use to list file information form from file systems, such as file size and number of files per table (recommended > 1 for blobstore).

...