The really interesting thing is that this indicates that Amazon really is copying the bits around to everyone's locker, rather than just saying "user x has rights to song y". Otherwise why would it take so long (12 hours+) just to give me access to these songs?
What is the reasoning behind not performing de-duplication of common files? It seems incredibly stupid to copy a popular song hundreds of thousands of times when a "single" canonical copy would do. Multiple people can stream music from a single file and people with large music collections would wait seconds instead of hours waiting for their songs to upload.
I'm curious if they really are doing that. It could lend some weight to their claims they don't need the music industries permission to launch a music service as they aren't streaming things, the users are accessing their own files.
I think they are. I bought the album and at first my locker had only a few of the songs marked as playable, a few minutes later, more were playable, and after maybe 5 minutes they were all playable. To me, the best explanation for this behavior was that it was taking some time for Amazon to copy all my purchased songs over to my locker.
ps: I'm not a big Lady Gaga fan, but for 99 cents I figured I'd give it a whirl.
...and is the assumption that the Amazon music service uses AWS or something like it, and the speed of EBS or something like it is too slow at this scale for the songs to be instantly playable?
Amazon tags the mp3 you bought from us with your customer ID. I suppose this is why each customer gets a very slightly different file in this case too.
If it would be only a tag, the web server could be modified to deliver most of the song from the same file (at least I'd do that). However if they do "watermarking" in audio data, there's no escape for them, every purchase is really demanding on CPU's.
