No sure if it is too early to ask, what is design target for max HDD/SSD capacity per Filecoin Node ? Or what is design target for data recovery in case one 10T/node data loss (for n-replications or Erasure Code schemes respectively)
Currently, there is no maximum implemented in the code, but we’re considering adding a cap somewhere around 1 petabyte in the future.
So what do you expect recovery time for 1PB loss (other participating nodes takes over all pieces of data in case one 1PB node complete down/loss?
In current data center solution, automatic data recovery is key factor to measure the performance of the storage cluster, it usually takes a few fours to several days for one full-loaded HDD e.g. 8TB data recovery.
This depends on a number of factors. If we’re talking about a failure with a single miner that is still online, that miner can implement their own solutions for backup and recovery, as Filecoin sectors are stored on their disk in a manner that’s totally customizable for them.
If we’re talking about an entire miner falling offline, it depends on the client’s choice of replication. Clients decide how many times they want their data replicated and which miners to store that replication with, so depending on what they choose, recovering a lost PB could be relatively quick or totally impossible.
other participating nodes takes over all pieces of data in case one 1PB node complete down/loss
I think there may be some confusion here around how Filecoin works. Nodes do not automatically handle failures from other nodes. In the event of a miner failing, a client could potentially download their data from other miners they’ve stored it with then make a deal with a new miner to store that data, but none of this is automatic in the protocol of Filecoin itself. It’s up to clients to devise their own replication schemes.