Why amazon s3 cli copy much faster than java sdk in the same bucket?

Question

I have a file in s3 bucket say like s3://prefix/file1 with size 2G I tried to copy it to s3://prefix/file2

The CLI I used is aws s3 cp s3://prefix/file1 s3://prefix/file2

The sdk usage is amazonS3.copyObject("prefix","file1","prefix", "file2")

CLI method took ~15s while sdk method took ~30s.

Answer 1

Try to use the TransferManager, like in this "old" blog post of AWS: https://aws.amazon.com/de/blogs/developer/using-transfer-manager-to-copy-amazon-s3-objects/

I'm not very familiar with JAVA programming, but in several applications I've been involved in migrating/optimizing, switching from the "normal" SDK methods to operations using TransferManager with Java has sometimes helped improve performance.

Maybe this could also improve the performance for you.