Peter Ebert created HDFS-14365:
----------------------------------
Summary: Filter block locations on storage type
Key: HDFS-14365
URL: https://issues.apache.org/jira/browse/HDFS-14365
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Peter Ebert
When using a policy like ONE_SSD, it would be advantageous to control what
storage types are accessed for a job. For example: low priority ETL job could
read from HDD replicas, high priority user queries could read from SSD.
To enable this the APIs would need to be able to filter based on storage types
for block locations, or for file open preferring to read from specific storage
types failing to others if none are available. Specifically thinking of
.getFileBlockLocations() and .open(). Adding an optional parameter to each
like: preferredStorageTypes=SSD
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
