EC Scheme Question

[NetApp111]

I am trying to better understand my erasure coding redundancy policies and what would happen if there was a failure within the StorageGRID. 

At one site I have the 4+2 EC scheme used at a site with 10 storage grid nodes. This site has 500TB of space total split between the 10 nodes. Does this mean that I can loose 2 full nodes and will still not have any data loss?  

On the other hand, we have a second site (different NetApp) with the 4+1 scheme at a site with 6 storage grid nodes. This site has much more space split between the 6 nodes.  Does this still mean that I can loose 2 full nodes and will still not have any data loss?  

Or how can I interpret my data redundancy policies?

[aronk · Accepted solution]

EC 4+2 means you need 4 out of 6 chunks of the object or parity to retrieve the object. This means in a worst case scenario you could withstand 2 nodes failing and still retrieve your objects.  The more nodes you have, the more chances you have that you could loose more than 2 nodes and still have the minimum 4 chunks of every object, but there is no guarantee.

With EC4+1, you still require 4 chunks of the object or parity to retrieve the object. With only 1 parity chunk, you are protected against loosing 1 node for sure, and again depending on the number of nodes you might be able to survive more than 1 node data loss but no guarantee. 

[elementx]

Just FYI, we have a StorageGRID area for SG questions here:

https://community.netapp.com/t5/Object-Storage/bd-p/object-storage