I’m ssh ing back to home I wouldn’t dare try that on the ships connection
I’m under the impression take the current total amount of data movement is quite low still, but I may be wrong. Do you, or anyone else, happen to have some metrics?
There is about 500kbps up and down traffic for each node multiply that by the network size of 74000 nodes. Then there is a pretty large amount of data flowing around the network.
But yes would be better with more people uploading and downloading to put more load on the network from users.
What part of that is from up- and downloads, and what is from the base level chatting of the nodes?
If you figure that out let me know
Do I understand correctly?
Attos that are spent on uploads cannot be counted for the leaderboards?
my understanding is its open season on uploads place is calculated from payments for uploads from the official up loaders only.
Qualifying data payments
The leaderboard ranking is based on the weekly earnings from the data uploads made by MaidSafe uploaders and our official partners. The wallet addresses of these uploaders will be published with the leaderboard for transparency and verification.
Why don't other uploads count?
It's reasonable to ask why all uploads to the network don't count towards these points tallies? This is for a few of reasons:
We don't want to introduce a loophole where the leaderboard can be gamed by people paying themselves through the smart-contract, faking data-payments. Publishing the uploader addresses gives transparency to everyone.
We also want the latitude to distribute tokens for people to use the network, and upload their own data. This allows us to do that fairly, for you it means you ca try the network without feeling like you're missing out.
We have more rewards and incentives progams to come 😉
Can I use the attos I earn to upload data?
Yes! You don't need to hold your earnings in your wallet. You are free to move them, or use them to upload your own data to the network should you choose to. This won't affect your place on the leaderboard, nor the rewards you'll earn, and you don't need to hold the tokens to get your rewards either.
At the moment you'll need to use the CLI tools to upload, but we've got a new app coming soon to make it easy for anyone to give the network a try, no experience neccessary!
Attos earned from uploads from non MaidSafe uploaders will NOT count (gaming rationale)
CC @aatonnomicc
So no uploads from me then
Which I think is a waste and we should be working this network just as hard as we can.
Perhaps you could share with us just how much data is being uploaded by Maidsafe and partners? Cos right now Im not convinced its enough. Yes Im earning attos - into three seperate wallets which is yet another distortion of how things will be when this thing is really real. Right now Arbiscan says there are >8.5k holders. How many actual individuals though?
Hows about we stop all incentives for a week, let folk use the attos that Autonomi knows we have already earned to do uploads and stress the network NOW - not when it’s “live” and we make a public arse of it cos we didnt try hard enough to stress/break it?
I don’t follow, why no uploads for you?
Why stop incentives to upload, we can upload now. Eli5 me.
and aside from the sheer joy of seeing how far we can push this network, there is no point, no one will benefit from my upload
As before, incentivising folk to participate distorts what we can expect to see in the wild. But I do appreciate that somehow we need to get more folk to participate. And until I have a better solution, I should probably stop moaning.
Still think we should all be trying to wring the last bit of performance out of this network NOW and if we have to fail, we fail NOW and can at least attempt to sort it, not later when it may be a lot more public and more difficult to sort. And we disappoint these hard-won partners amongst others.
Months ago (before you were on the team mainly) multiple times (first in April I think), we developers were told “don’t worry” because there was going to be a way for us to test out apps without having to earn tokens first.
It wasn’t spelled out what was intended, but we’ve had no help at all from Autonomy. And even local test network doesn’t work any more since ERC20.
Luckily I had help (many tens of thousands of nanos and now thousands of attos and also gas) from a handful of community members who helped out by donating their tokens and losing out on rewards by doing so.
That’s still the case. Why? Is any help planned at all or was it just empty promises?
If help is planned, what and when?
The current scheme is rapidly reducing incentives for the very people we were - if the fundamentals still mean anything - meant to be prioritising. This from me on Discord:
I’m only able to run a handful of nodes (down from ~60 on same equipment 3 weeks ago) and am earning far less attos, rank down from ~70 to >800 and rewards down from 100 to 8.
That’s with me just doing the same thing. This doesn’t bother me too much as I’m just running nodes to keep abreast of the project while I spend time on apps. But it doesn’t seem a good look when we are supposed to be attracting small players who aren’t going to spend time optimising things.
Does this not bother anyone else? We are seven weeks from launch and it’s only maidsafe and a few community members uploading/downloading. Surely this needs to be tested on a larger scale to see how it impacts the network?
Some fears may be allayed if Maidsafe/Autonomi would tell us just how much data they are uploading so we can get a handle on it.
I have no complaints about the qty of attos currently being earned, I’d just rather do something useful with them. And know a lot more about what earning rate relates to what overall upload rate.
I’d like to test hundreds of people trying to download the same file at the same time to see if everyone gets it, and all sending some attos to the same wallet to see if they all land.
Definitely need about 100 times the uploads AND 1000 times the downloading to get a realistic figure. As soon as that happens each node will be using a little more CPU and many people will lose a few/many nodes and we’ll see a better picture.
The error rate will rise dramatically (ie more volume of traffic due to retries) with more large chunks flowing rather than the vast majority being a few hundred byte protocol packets. Once data chunk traffic really exceeds protocol volume then it’ll be a real test of the network. At the moment “we are specifically asked not to just take larger node setups off line too fast”, “uploads are still slow”, “downloads almost non-existent” which adds up to treating the network with kids gloves.
The reason I say the error rate will rise with more usage is that the volume of traffic we see is mostly protocol traffic and the majority of it is one UDP packet in size back and forth adding up to retry volume very small. A chunk of any significant size is many udp packets in length and ALL packets have to arrive in good condition for the whole chunk to be considered valid, otherwise a retry of the whole chunk will occur. So while 1 in 1 million (to 100 million) UDP packets transfer through a switching/routing point in error, the error rate becomes significant once chunks are transferred.
EG for a 4MB chunk we have over 2,500 udp packets being transferred though at least 12 switch/routing points (2 only between nodes within datacentre like DO) the error rate figure used is increased by 32K. Thus the error rate is like 1 in 31 to 3100 chunks which for a setup with 100 nodes means you’ll see one chunk error every few time periods for an active network with uploading/churning/downloading occurring in volume. Of course this can be reduced with setting the QUIC window size down to 128K (1/32) which would reduce the error rate to 1 chunk in 1000 to 100K window size. The retry is the full window size sent again, max 4MB chunk currently. And if packets are dropped due to ISP router buffer size then its even worse.
Hopefully Dave (the easy to use app for uploading & downloading stuff) will sort this out and enable loads of people to upload & download to test things out & put some more stress on the network.
Sorry Dave is only going to make it easy and faster for shit to happen if what @neo says above is correct.
And he usually is.
The proper testing has not been done, nor does it look likely that it can be done in the next few weeks - which includes Xmas and New Year…
I know we have all been screaming for a Dave-like browser thing for many years but resources right now should be directed to proper testing.
Publishing a test schedule would help. But that might mean involving the community…
Can you please clarify this point?
Does it mean attos users spend on uploads won’t add to leaderboard positions for those who store the data, or that if you spend attos that you earned, you will lose leaderboard position by using your attos to upload.
I assume it’s the former, as if it’s the latter, it seems to contradict this: