According to the calculator, upload speed is the “golden goose” for pay. My ISP has recently bought out a new package option for double my current upload speed. If I were to upgrade to this, how long would it take the satellites to “notice” and give me more traffic (I’m in my 10th month)? (Huge assumptions about how the system chooses to prioritize data - so please correct me if anything is wrong)
1 Like
Sattelite dont know anything about your speed, and event dont whant to know. It wont give you more trafic. it only effect if client want download data, and faster you serve it more you can earn as if demand is bigger than you can upload, then some of pieces than can be downloaded from you will be dolnloaded from other.
Hit the breaks for a second. The earnings estimator on storj.io is useless. Chances are the upgrade won’t do anything for you at all. What speed do you currently have and what are you planning to upgrade to?
Please also check out this post.
I currently have 4.5TB stored and it doesnt need more than a 12Mbit/s upload connection.
1 Like
I have 11,2 TB stored and average egress is 720 KB/s. AVG Ingress is 1370 KB/s…
average is not that helpful for determining the needed bandwith because you might have 50% with no egress.
I was monitoring my docker network usage and most spikes don’t go over 13MBit/s so my assumption is that a 12MBit/s connection seems to be just fine at the moment and catches most spikes.
ingress shows spikes up to 40MBit/s.
1 Like
Sure, you are right. But how many spikes are there? Do they make more than 1% of the time?
Maybe it makes more sense to use Quantiles.
But if he looses only 1% of the egress „races“ during these spikes, it probably would not make sense to upgrade his bandwidth then.
about 30mb/s upload and I have 6TB in use. Wow that linked spreadsheet looks really accurate to my current earnings. For my input values it claims $280 for a year and I’m currently sitting at about $240 after 10 months (although my first 2 months was during surge pricing). Going by that, it looks like I’d need to increase my raw storage space a ton before faster up or down speeds would make a difference. Thanks for linking this, its really useful
Yes only looking at spikes is not helpful either in determining which connection speed is really neccessary.
I just assume that most people will have no trouble getting a 12MBit/s upload connection (at least in Germany you can’t really get lower speeds unless you have a bad connection). So even this low upload speed catches almost all spikes. So higher ones hardly give you any advantage.
The spikes depend on the client’s download bandwidth. So if a client downloads 35 pieces at a time and has a 500Mbit/s connection, the nodes can upload with 14Mbit/s on average. but that calculation is optimistic and optimal so real world will be lower.
1 Like
Thanks for that feedback, I’ve done my best to make it represent actual network activity, so it’s good to hear that it matches your experience quite well.
And as I expected you would almost certainly see almost no advantage of higher upload speed. It rarely peaks over 30mbps.
If you want to see how much you could possibly gain you can use the success rate script here.
For now you should ignore the upload numbers though as there is a loggin issue that impacts the success rates. But since you’re worried about YOUR upload speed, the success rates for download are the relevant ones. (Upload/download terminology in Storj is always from the perspective of the customer, so a customer download uses your upload)
Here’s my result.
If you’re close to 100% there is nothing left to gain from better upload speed.
I only have 83% on download. I also have 55% on upload. This might be due to the device being choked by the media centre I tried setting up recently (I’ve moved it off now)
Yeah, ignore the upload for now though, I have 26% upload success. It’s a logging issue, not an actual node issue.
With 83% there is something left to win, but at best you get 17% more. And I think you’re right that other optimizations may already get you that 17%. Try that first and see if there is anything left to win after that. That 17% is probably already not worth the upgrade to get higher upload speed though.
4 Likes