You can exercise it against regular, non-Glacier S3. Same flow; the only difference is that you skip one attribute on upload and one request/wait step on download.