I understand that IP filtering treats multiple nodes that are operating under a single public IP as a single node. But, does this only lead to degraded performance regarding ingress? If you simply accept worse performance until your various nodes have filled up with data, will you from that point function efficiently since most of your data traffic will be egress?
Well, your nodes get treated as one big node when ingress is concerned. Egress depends on the amount of data stores and what data is in the nodes (if the customers access it or not).
Basically, if you have 4 separate nodes on separate drives, but one IP, they will perform, combined, as a single node with the total storage space of the four nodes. However, running separate nodes on separate drives is easier (no need to set up RAID) and if one node gets disqualified for whatever reason, other nodes may not be.
As a side note, if you have 3 full nodes and a 4th one filling up, all behind the same IP, the 4th one will get as much ingress as it would if it was alone, because other ones won’t be selected anymore by satellites for ingress.
@Pentium100’s right it does not impact egress which is directly related to how much data you store.
Well, mainly at least: performances of your setup could impact that; see success rate:
So when we have filled up, say 4 out of 4 drives, we will not get any more ingress unless data is deleted (removed by customer) and new data flows in? Does this mean that we stop getting paid for ingress, and only get paid for future egress? Or, do we get paid for simply storing data as well?
You get no money for ingress. Only for storing data and for the egress.
Aha, then I think I understand it properly. Thanks all!