Why do the AWS docs instruct us to use the raw (uncompressed) file format? AWS supports other formats and the raw file format is extremely inefficient

[colbyn]

Why does the AWS docs instruct us to build our images using some "raw" file format?

Why not just use e.g. VMDK which is significantly smaller (e.g. from 1GB to ~100M)?

[colbyn]

I don't know if this is related. But for some reason I couldn't get LinuxKit to upload e.g. VMDK encoded files to AWS, it just blocks (seemingly forever).

I have a suspicion it's due to this line, that seems to hardcode a given format (even though AWS supports other formats and the raw file format is extremely inefficient). I.e.

importParams := &ec2.ImportSnapshotInput{
    Description: aws.String(fmt.Sprintf("LinuxKit: %s", name)),
    DiskContainer: &ec2.SnapshotDiskContainer{
        Description: aws.String(fmt.Sprintf("LinuxKit: %s disk", name)),
        Format:      aws.String("raw"),
        UserBucket: &ec2.UserBucket{
            S3Bucket: aws.String(bucket),
            S3Key:    aws.String(dst),
        },
    },
}

Anyway for now, I created a simple tool that automatically uploads the given image to AWS.

E.g.

$ ami-uploader upload --bucket my-s3-bucket --image ./my-source-image.vmdk --name my-aws-ami