Bucket quotas in combination with compression

[kowalczyk-jan]

How does S3/RustFS handle bucket quotas for compressed data? Is this standardized in S3 or is this up to the implementation?

I think currently in RustFS the bucket size is calculated using the uncompressed file sizes, not the actual size of the compressed data stored on disk. Depending on the usecase, it might be more interesting to respect the quota for compressed data size, or am I misunderstanding resposibilities/best practices of this feature?

Can/Should there be an option in the future to select if the quota is applied to the uncompressed file size or to the compressed file size as it is present on the underlying storage? I'm not knowledgeable about S3 enough to know if this is defined, or if compression is defined/handled by S3 at all vs. left to the implementation to handle.