The concept of your information “shards” scattered on a super-massive hard-drive sounds intriguing.
Pieter Glasbergen {G888}
1

Pieter Glasbergen {G888}, you highlight an important point and that is how ShatteredGlass is intended to operate.

I used the concept of a single (universal) storage container only as an extreme stance to drive the design of the Sharding system where there is no centralized account management and no centralized namespace management (namespace anarchy) yet the billions of Shards from millions of distributed and uncooperative users must be able to be stored in a single location without collision. That is, there is no Universal Information Identifier (UII) collision or Shard collision. I did not mean to imply the system must store all the Shards in a single container. On the contrary, the Shards can and should be scattered over multiple storage locations.

In ShatteredGlass Part 3 (yet to be written) I will describe how Information Dispersal Algorithms (erasure coding) for data redundancy is used to allow a certain percentage of Shard loss as the Shards are distributed/scattered across multiple storage locations. I introduced the use of Information Dispersal Algorithms in the ShatteredGlass GitHub readme, point 3 (it just won’t be visible until the Part 3 Medium article is written).

Combining Shards scattered across multiple storage locations with Information Dispersal Algorithms allows failure of a percentage of storage nodes or loss of a percentage of Shards yet the system is still able to reconstruct the original Information, much like how the routing system of the Internet can reroute packets around failed router nodes.

Show your support

Clapping shows how much you appreciated Daniel Lanovaz’s story.