# Amazon EMR 7.5.0 - Hadoop release notes
<a name="Hadoop-release-history-750"></a>

## Amazon EMR 7.5.0 - Hadoop changes
<a name="Hadoop-release-history-750-changes"></a>


| Type | Description | 
| --- | --- | 
| Bug Fix | Commented out fs.file.impl to empty value. | 
| Backport |  [HADOOP-19286](https://issues.apache.org/jira/browse/HADOOP-19286): Support S3A cross region access when S3 region/endpoint is set | 
| Improvement | Automatic S3 region configuration setting for S3A connector on EMR-EC2 | 
| Improvement | Reduce the number of HeadObject calls in S3A | 

With the release of Amazon EMR 7.5, Spark's S3A connector demonstrates read performance comparable to EMRFS, as evidenced by benchmarks using a 3TB TPC-DS parquet dataset.

## Amazon EMR 7.5.0 - Hadoop features
<a name="Hadoop-release-history-750-features"></a>
+ S3 region configuration `fs.s3a.endpoint.region` is automatically set to the region where the EMR cluster is launched with S3A connector for EMR-EC2 deployment. 
+ Amazon S3 cross-bucket region access is enabled by default for the S3A connector. It can be modified by setting `fs.s3a.cross.region.access.enabled={{true or false}}`.