Re: I know what they want to imply by "100 %"
Responding for Scality - This is achievable. Traditional storage architectures for sure cannot achieve this. However, since scality (and others) are distributed storage systems and automatically rebalance data across the cluster after failure, you automatically avoid failure points. The system can be configured with many protection copies for durability using distributed erasure coding - this means that you can protect at any one time ffrom many disk and server failures. The 3rd element which then makes 100% possible is that you need to be monitoring the system and fixing dead disks/servers which may have gone wrong. This is what HALO then brings to the picture. With a system designed for any failure scenario and continuous monitoring then 100% availability is definitely achieveable.
Learn more about the scality ring architecture here http://storage.scality.com/white-paper-scality-technical-wp.html