I’ve just started running a node, and I saw in the documentation that it says that upload speed makes a lot of difference of which nodes get prioritised, however I just wanted to check if that is actually the case before I go upgrading my provider to faster speeds
I’ve gone through my logs and it seems that my average data upload is 0.28mb
My current upload speed is a tiny 6mb, but even at that speed, that means that the average upload would only take 0.047 seconds.
Do you have a lot of “upload canceled” or “download canceled” log entries right now? If so, upgrading your connection speed MIGHT help.
Connection (and CPU, HDD) speed matters up to a point - you do not want your node to be one of the last ones to finish (because it will get canceled), but there is not much point in being first by beating the second place node by a lot.
I would say that’s enough for a 6-8TB node. I would work with QOS so that you prioritize the node backwards and your Internet does not become so slow when surfing.
Maybe improve the internet when the node gets full, but I wouldn’t do that until then.
EDIT: I also only have 10Mbits upload. This is what it looks like at 11TB for the past 30 days.
16TB node monthly avg 1000/500Mbit fiber the graph is in Mbyte/s so 1200k is 10Mbit
so in theory 10Mbit up could become a limitation and will certain eventually reduce the peaks and extend the customer download latency.
but for most moderate sized nodes i would think 10mbit to be fine.
this is the go to solution for most of us.
generally with near perfect performance one will get about 99%+ on everything
but downloads and uploads can often be in the 97% to even 92% ranges depending on lots of factors
if you need some actual numbers you can reference them here.
Some things to keep in mind though. Any graph that displays peak transfer speeds measures that over a specific time. Pieces being transferred are pretty small. So if a transfer doesn’t last for that monitoring time window it could be that actual peak speeds were higher than reported. Those success rates are a decent indicator, but unfortunately the logging they are based on is not as reliable as one might hope. Successful transfers in the log may not always end up with your node actually getting to keep the data. There is a long tail cancellation of uploads of 30 out of 110 pieces. Yet most report success rates in the high 90’s%. Now it’s possible that it’s just the best node operators who monitor and post that stuff and I’m sure that’s part of it. But I’m afraid there is probably also a sign that actual success rates may be lower than the logs report.
For reference, this is what mine currently look like across 3 nodes. And it seems just a little bit too good to be true.
So Any result below say 95% may be an indication that there is room for improvement already. If you’re not sure, just post your numbers in the comparison chat and ask for feedback.
