According to the document, aws s3 sync will compare the content of all the 2TB files even when we just added 5GB files. It would be good to avoid checking the 2TB files.
Even though it is still not ideal, I think Fred had partially optimized it, because previously we had to downloading the 2TB files before the updating, which is worse.
We use
aws s3 sync
to upload deb packages from EBS to S3 https://github.com/hhvm/packaging/blob/6d301967b15491b4c504947ec9bf0215ffb1a803/bin/update-debianish-repo#L49-L50According to the document,
aws s3 sync
will compare the content of all the 2TB files even when we just added 5GB files. It would be good to avoid checking the 2TB files.