[vgb2k11]

>The webpage also says, "Redundancy is achieved through the use of erasure codes so that your data can always be recovered even in the event of large network outages."

>Does this means files can't be lost, as long as you keep paying your bill?

The white-paper mentions:

>"ORC will soon implement client-side Reed-Solomon erasure coding (Plank (1996)). Erasure coding algorithms break a file into k shards, and programmatically create m parity shards, giving a total of k + m = n shards".

So at first glance its seems like the usenet parchive/PAR2 redundancy methodology, but storing the parity shards locally (client side). Well that's my interpretation of this section of the white-paper anyways.

So in short: it certainly doesn't mean that the files "can't be lost", but it means the owner of the files can rebuild the files using parity shards from client side in that case that network outages affect file availability.