I try to achieve the max performance in download/upload from/to S3 compatible object storage.
I have available 4 Gb/s network connection and a block volume of 900MB/s capable (180GB RAM and 12 CPU Cores), but I reach only 360MB/s in read and around 280MB/s in download.
--s3-chunk-size SizeSuffix Chunk size to use for uploading (default 5Mi)
--s3-disable-checksum Don't store MD5 checksum with object metadata
--s3-upload-concurrency int Concurrency for multipart uploads (default 4)
Also try this - we've seen big speedups with this recently
--s3-disable-http2 Disable usage of http2 for S3 backends
I suggest you carefully read all the posts in this recent thread:
In it you will find an entire walkthrough of a systematic approach to tune rclone up/download parameters.
You may not need to tune the S3 specific parameters once the general optimization is completed, the bottlenecks are typically in more general areas (think --transfers and --checkers).
This was one of the learnings from the thread linked above
The best performance that I can reach in upload doesn't exceed 370/380MB/s
The download it's more problematic .... start above 300MB/s and after when ETA go to 0s the speed decrease around 0/b. The hdd is reading (???? why normally has to write...)...
At the end the download consume the double of time than the upload.....