Probably almost all of it. Right now, almost all data seems to be test data because developers are mostly trying how the system works etc but are not moving big amounts of data.
I have one full node with 900GB and it doesnt get any downloads.
My 2nd node with 500GB is almost full too and my 3rd node already has 2/2.8TB so maybe my ingress will stop this year too
But since STORJ plans to be production ready in the first few month of 2020 it shouldn’t be too long before some test data gets erased (I guess).
Yeah from what I can tell the vast majority is test data. Which also means that the current effect of full nodes not getting egress is not at all representative for normal production use. In fact, in normal production use, most like the more data you hold, the more downloads you see.
I wouldn’t worry to much about the short term, keep the nodes running and stable and you’ll be a step ahead by already having good reputation when more actual customer data starts rolling in.