The median reachable Bitcoin node's upstream bandwidth dropped by 39% year over year to 6.6 megabits per second. This is both baffling and concerning.
Discussion
define "upstream" in a peer-to-peer system
Because of knots?
Nodes with low upstream capacity might be poorer connected or under higher load or have other bottlenecks. Running 50 nodes on one VPS is not the same as 1 node on a barebone.
How do you figure? There aren't significant differences vs. Core in the relevant part of the code.
I don’t figure anything. I asked a question.
Csn you explain a bit more
It just means that the speed at which you can receive block / transaction data from the average peer node has been going down.
And how is this measured? Maybe it says more about the systems that are doing the measuring, than it does any the systems being measured?
According to bitnodes: Network speed of a reachable Bitcoin node is measured once per day by downloading a sample block from the node. The aggregated data includes only reachable IPv4/IPv6 nodes that do return block data at the time of measurement.
Interesting. Thanks for that. I guess I should have checked myself 🙂
Do they consider only archival nodes, and request a random block uniformly going back to the genesis block? Or just from the most recent few hundred blocks?
I ask because this might be basically about increased size of the chain. More blocks means the node is IO-bound while searching for an old block
Also, are recent blocks bigger? My BlockStats Explorer suggests that blocks are getting bigger in the last year or two
maybe they are not public?
More Tor-only nodes post Samourai arrest?
That was my first thought but bitnodes says it doesn't include bandwidth measurements for Tor nodes.
Also, the ratio of Tor to clearnet nodes didn't change meaningfully last year.
Is this measurement such that increased bandwidth demand per node is a possible cause?
Hmmm perhaps, though I haven't observed increased bandwidth usage on my Statoshi node.
congrats on 400 days uptime
