Skip to content

"rsync for cloud storage" - Google Drive, S3, Dropbox, Backblaze B2, One Drive, Swift, Hubic, Wasabi, Google Cloud Storage, Azure Blob, Azure Files, Yandex Files

docs: add section for improving performance for s3

Pawel Paluchada70088b11e6
diff --git a/docs/content/s3.md b/docs/content/s3.md
index db52f5c26..9031e57fe 100644
--- a/docs/content/s3.md
+++ b/docs/content/s3.md
@@ -401,6 +401,38 @@ there for more details.
Setting this flag increases the chance for undetected upload failures.
+### Increasing performance
+
+#### Using server-side copy
+
+If you are copying objects between S3 buckets in the same region, you should
+use server-side copy.
+This is much faster than downloading and re-uploading the objects, as no data is transferred.
+
+For rclone to use server-side copy, you must use the same remote for the source and destination.
+
+ rclone copy s3:source-bucket s3:destination-bucket
+
+When using server-side copy, the performance is limited by the rate at which rclone issues
+API requests to S3.
+See below for how to increase the number of API requests rclone makes.
+
+#### Increasing the rate of API requests
+
+You can increase the rate of API requests to S3 by increasing the parallelism using `--transfers` and `--checkers`
+options.
+
+Rclone uses a very conservative defaults for these settings, as not all providers support high rates of requests.
+Depending on your provider, you can increase significantly the number of transfers and checkers.
+
+For example, with AWS S3, if you can increase the number of checkers to values like 200.
+If you are doing a server-side copy, you can also increase the number of transfers to 200.
+
+ rclone sync --transfers 200 --checkers 200 --checksum s3:source-bucket s3:destination-bucket
+
+You will need to experiment with these values to find the optimal settings for your setup.
+
+
### Versions
When bucket versioning is enabled (this can be done with rclone with