I have 2 new nodes started, one half a month, one a month old. I will let only one sat on one of these and report back what I find in the next months. My guess is in the long run, the ingress will be smaller on it, but will fill up with data than can provide more engress, which pays out more. But will see. Maybe is not a big difference.
I have one machine with 2 nodes, the only one with IronWolfs, upload available 150 mbps, filled with 5TB and 1.5 TB. The older one has droped online scores to 98.xx. I donât know if is caused by not serving all the engress requests, but maybe droping some nodes will increase my score with 1 sat.
About the effect on decentralisation, if the whitelisting of 1 sat will become a general trend, Storj can disable it at any time. Itâs up to Storj Labs to allow whitelistings or not. For the moment they allow it, so we are playing by the rules. If this will be disabled, the nodes that exited some sats will be again in the vetting for those sats, for 1 month.
I would definitely like to see the difference. Thanks for using your hardware for profitable science!
BTW⌠how do you Graceful exit only some satelites, and not all at once? I didnât checked the procedure, but I think there are steps only for total GE⌠Or is just best to whitelist the one you want in the config, and thats it?
You list all satellites you want to exit. You can list any number.
wrong assumption. The customers sends data directly to your node not trough a satellite, so practically doesnât matter where is satellite located, only important where the customer is located (relatively to your node). But there is no way to figure that out exactly.
Wouldnât be illogical to assume that the customer is likely to use a satellite closest to them though. Of course this wonât apply to use cases where Storj is used for global distribution.
That said, I am also not convinced that distance to the customer is the determining factor in earnings per TB. Itâs possible that the spread of use cases is just different among different satellites.
My earnings per TB in October (based on how much data was stored by the satellite on 2022-11-01):
| Satellite | TB | USD | USD/TB |
|---|---|---|---|
| ap1 | 1.86 | 3.5 | 1.88 |
| eu1 | 2.93 | 6.68 | 2.28 |
| europe-north-1 | 8.79 | 15.45 | 1.76 |
| saltlake | 7.52 | 32.89 | 4.38 |
| us1 | 1.93 | 13.18 | 6.82 |
| us2 | 0.02 | 0.04 | 1.79 |
| total | 23.05 | 71.74 | 3.11 |
These should be decimal terabytes.
By far the best $/TB was US1, even though I am in Lithuania. Saltlake gave me the most money in total though.
Thatâs what I expected to see. Use case differentiation, not distance.
You wonât receive more traffic from whitelisted satellites if you blacklist some other. Given the numbers shown (and which I also observe on my central european nodes), as long as you have spare resources to handle less profitable satellites, you should try to accept all traffic. Only when you are limited by storage, ingress, egress, IOPS, etc., then it might be worth thinking of blacklisting.
There are some other ways to limit traffic as well. For example, if youâre limited by IOPS, you can deny small uploads. This probably requires some technique for inspecting incoming packets before they reach the node, but should be quite possible.
If you can recognise geographical areas from which data is less profitable (either because thereâs a big risk of losing races, or just it has been identified that data received from that area is very short-lived and/or never has egrees), you can deny uploads from thereâagain, with some packet inspection before the connection reaches the node.
Lots of choices. Not that I recommend any them, it should usually be easier and more profitable to just expand resources and handle all traffic. Besides, Iâd fully expect Storj to implement some kind of PUT_AUDIT request if these techniques become popular 
That would probably require modifying the source. Instead of just denying small uploads, it would be better to put in sleep(10) to pretty much guarantee the node loses the race, but it would still allow the customer to upload the file if other nodes fail.
just make GE from sattelite you dont want thats all.
I also checked now the income from satellites and it seems that no mather where you are, the busiest sats are Saltlake and US1, and they give the most income. I was presuming initialy that all sats are equaly busy, without verifing them. Maybe in a perfect world were all sats see the same traffic, my scenario whould be realistic. But for now, we should keep them all whitelisted.
It is because most storj client are in US for now, I think so.