Yep, and still a ultra low percentage of the world’s people with computers. What maybe a few hundred/thousand people running hundreds of GPUs each. In fact the miniscule %age shows that most of the world do not think the way you are saying
But crypto people think exactly like that. Chances are they’ll be the majority of nodes in the early years, and the marketing focus shows the team thinks so too.
Check out the Dev Forum
And now we are getting into reality. It will be a very small %age of the world doing as you think. Crypto world is still very small and even smaller are the ones who will do this. Many in the crypto world are not tech people either. Just in it because they want to make a quick buck. Node running is not a super profitable business so yea, my opinion is that there will be maybe thousands who will just keep buying drives to make more but a thousand times the people who will not as the world adopts Autonomi
And this is where our opinions differ. To me, adoption is not the same as everyone running nodes. My opinion at the moment is that when mass adoption comes it won’t be worth the click to start a node and the storage space will be so cheap that the mass person won’t even have to buy tokens to use Autonomi, everything will be outsourced at the apps level.
Check out the Dev Forum
I suspect some will add disks, others would rather nodes were stopped/cleaned up. Both solve the problem.
The questions is, is having people dedicate more CPU than storage good or bad for the network?
1 Like
Same, but only 16 nodes.
This though raises another concern about the readiness of the network because as it has scaled up, it has become ever more empty and we have no testing of how churn of much fuller nodes will affect a network at these scales or larger.
If nodes continue to join at the current or higher rates, and remain so far ahead of storage we may not know how this will behave as nodes fill up for months, and that could be a nasty surprise that hardly anyone will be expecting.
10 Likes
Great question, several answers for sure lurking in the shadows of this community of busy beavers, the question really is will they share… 
I don’t know the answer because Antode code has been in a constant change/evolution, so it makes no sense to put an Application Monitoring Tool on them to profile their activity (ie New Relic, App Dynamic Dynatrace or similar might be able to reveal a bit of truth as to what is going on to profile antnode i/o and resource use behaviour over time and in different primary, secondary and exception use cases, more accurately)
Right now its a game of darts and the board is revolving at various speeds…randomly changing…, tough to invest resources when that is going on…
2 Likes
I dont see a differenxe between being bottlenecked by storage, cpu or memory.
If you run out of it: run less nodes
6 Likes
Which people probably have a hard time to recognise atm (it their bandwidth is the bottleneck)
3 Likes
yep thats the fscky one.
Would it be possible to make a benchmarktool that can find out the nat table size ?
Before the last update with 500 nodes it was around 280 000, currently with 1500 nodes it is around:
sudo cat /proc/sys/net/netfilter/nf_conntrack_count
87570
Check out the Dev Forum
2 Likes