I see that there is possible to reduce your processing fees, by outsourcing auditing and reconstruction of files, to special nodes with contract, I would even agree, that storj inspect this pc with remote connection, as i have dedicated machins for storj. This trafic can be some 5$ tb.
As i know Storj use Hetznet for 20$ for TB of Egress?
4 Likes
it also will be good if storj publish some operational cost for runing network, i mean servers, auditing, recreation of files when it drop below threadshold. As i understan this costs are much bigger that those that you pay to storagenode operators. And it is better to optimise them first, and make them possible to make additional profit to storagenode operators than paying them to server companies.
4 Likes
That question did not receive any reaction so far.
So maybe an additional thought on that: I believe the current model is in favor of egress than pure storage. This could mean that SNOs are incentivized to get rid of data that does not produce (much) egress like backups. This could mean that if you run a node that is full and cannot be expanded for some reasons, you could be tempted to rather do a graceful exit (which additionally returns your held amount) and start all over to be able to receive more valuable egress pieces again than keeping your existing node online. This does not sound healthy for the network altogether.
@BrightSilence
I have no data on this but I believe you could tell. Is this something we see? That a full node has a sharp drop in earnings as no new egress pieces will land on that node?
I see that old nodes 1-3 years old make less egress, it can be because they are full of test data.
New data coming today give more egress.
Yes, I believe so. But by far the biggest impact is whether you still receive ingress. About 15% of ingress gets egressed in the same month. The moment the node fills up, that stops. Then on top of that you see about 6.5% egress on statically stored data. Now I’m sure that average still has some preference for newer data, but most my nodes show that older data still gets accessed. at maybe around 3% egress vs stored.
Some of this is inconsistent egress on test data, we’ve seen egress drop to basically 0 on the test satellites from time to time. Storj tries to replicate average egress, but I guess that process gets interrupted some time.
More importantly though, it’s just kind of natural that the older data gets, the less it gets used. This does lead to a slight inverse incentive. But in my opinion not enough to be worth it to start a new node and exit the old one. The time investment required to fill that back up would negate any gains you would get from that. If possible, just ensure you always have free space so you never miss out on that juicy new data. 
Ps. The split between egress on static storage and egress on ingress has been part of the earnings estimator for a while, so you can see the drop effect if you set it to for example 5TB.
Tying that back to the economic model. It would be good if Storj incentivized storing data for longer. Instead of paying a static amount per TB, why not increase that amount slowly over time, so that more loyal long term nodes get rewarded for reliably storing data long term.
6 Likes
I thought about that too. But simply node age is not a good measure I think. it might be to broad and too easy to gamble. An additional reward for long term storage should be nailed down to the age of the pieces that a node is holding.
My baseline is, a SNO should not have to even think about it, what kind of piece has more value for him.
Oh yeah, agreed. That’s what I meant to suggest in the first place. New pieces should still see more egress, so no need to boost static storage income on those anyway. This would really help prevent SNO’s from being tempted to somehow remove old data or exit satellites.
1 Like
Taking this into account, maybe even $1/TB for storage for new pieces would work, if it goes up to $1.5 for old pieces over time.
Might be worth to think about something like that.
2 Likes
I would prefer that we have a separation of concerns, i.e.
SNOs add value by storing data → incur the cost of storing data → get paid for storing data
SNOs add value by sending data → incur the cost of sending data → get paid for sending data
The cost and the value added for each activity then pass on to customers, separately. That way, the incentives of SNOs and value of service to customers can be well aligned. That’s the ideal.
As an aside, long-time rarely used data sounds like a cold storage. Cold storage is usually less expensive to store but more expensive to retrieve.
It would be hard to emulate for any rational reason.
I don’t think many SNO’s will try to weed out old data versus active data. I think everyone just wants more data, warm/cold, they just want more of it to fill their drives.
1 Like
I’d say it depends if you have space available and/or can expand. This is not necessarily the case.
I guess I don’t understand what you would do as a node operator here. Remove data that is cold? You would fail the checks and become disqualified. Gracefully exit an old node because it has become stale and then wait X number of months to refill it with new data? I don’t think that would be profitable.
How do you see removing cold data in favor of warm/active?
2 Likes
Use 3 nodes, once your available space is full, exit the oldest one and start a new one. Rinse repeat.
It’s not worth it currently, but we only have 3 years of data atm. What happens when some data is 5+ years old. Some old full nodes may never see egress again. I could buy a new HDD or just exit a node and get fresh ingress. If that doubles or triples egress, that might become worth it.
4 Likes
Plus I get the remaining 50% of my held amount back on graceful exit.
Yeah, I know. Just try to give a tangent example of why paying SNOs more for storing old data might not make sense.
Could you elaborate? I agree but wanna know what’s your take.
If node operators try to get rid of old data instead of keeping it, it might make sense to pay more.
Even more if node operators choose not to do a graceful exit and the old data needs to be repaired.
1 Like
I mainly see graceful exit as it gives the incentive of returning the remaining half of the held amount.
As I understand it, vetting will become fast again in the future, so a node operator will not have to wait many months until he receives full ingress again with the incentive the at current prices he receives 20$ per TB instead of (or even additionally to) 1.5$ per TB.