[SPARK-10340] Use S3 bulk listing for S3-backed Hive tables

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 1.4.1, 1.5.0
• Component/s: SQL
• Assignee: Cheolsoo Park
• Reporter: Cheolsoo Park
• Created: 28/Aug/15 20:24
• Updated: 08/Apr/16 11:15
• Resolved: 08/Apr/16 11:15

Description

AWS S3 provides bulk listing API. It takes the common prefix of all input paths as a parameter and returns all the objects whose prefixes start with the common prefix in blocks of 1000.

Since SPARK-9926 allow us to list multiple partitions all together, we can significantly speed up input split calculation using S3 bulk listing. This optimization is particularly useful for queries like select * from partitioned_table limit 10.

This is a common optimization for S3. For eg, here is a blog post from Qubole on this topic.

Issue Links

• depends upon: SPARK-9926 Parallelize file listing for partitioned Hive table
• duplicates: SPARK-9926 Parallelize file listing for partitioned Hive table
• is superceded by: HADOOP-12810 FileSystem#listLocatedStatus causes unnecessary RPC calls
• links to: [Github] Pull Request #8512 (piaozhexiu)