I became a SNO just around three weeks ago as I wanted to give this a shot. I got a few used HDD’s as well as a Raspberry Pi 3 B which is running 24/7 in my home network and downloading/uploading data. So far everything is good. I’ve also added automatic restarts after failed healthchecks - and it’s working quite robust currently.
However, I notice, that the actually used bandwidth is way below the limit. I’ve 50Mbits/s download and 10Mbits/s upload from my inet provider. But after multiple improvements, storj only uses around 135 kbits/s ingress and 90 kbits/s egress.
The graph shows the combined used network bandwidth of all three nodes (behind the same Pi).
Since I have DS-Lite connection, I cannot expose my Raspberry Pi 3 directly to the internet (no public IPv4 addr). Hence, I have a few low spec cloud servers (1GB RAM, 15GB disk) running in “geographically nearby” datacenters. These have public IPv4 addresses and forward the incoming connections to my personal Raspberry PI via a SSH TAP Tunnel. All three public servers are in different IP subnets (they don’t share the last octet). The graph indeed shows, how the traffic increases after 19th and 22nd January. Before that, they were all using the same proxy server.
I’ve three nodes running and I checked all kinds of bottlenecks:
HDD’s are fast enough → not bottleneck
Pi 3 B is fast enough and the 4 cores are underutilised → not bottleneck
I’ve checked the throughput over the SSH TAP tunnel on TCP und UDP. It’s as high as my inet-provider allows. → not bottleneck
I’ve checked the latency, and it’s quite good. → not bottleneck
I’m living quite centerally in Europe (many nodes nearby according to https://storjnet.info/) → being too far from where stuff happens is not the bottleneck
TCP and QUIC is all configured properly and work robustly → not bottleneck
I checked the subnet neighbors. Only my third node has another node as a neighbor → it doesn’t seem to be the main bottlenck
used bandwidth is higher/lower during day-night cycle, but it doesn’t really corellate with when I’m awake and doing stuff → not botleneck
With all this in mind, I’m asking myself, why my used bandwidth is 90-135 kbits/s. In the forums, I’ve heard of people getting quite a bit above 1 Mbits/s.
What can I do, to improve my setup?
I’ve checked the vetting status - and all the nodes are around 20%-40%, since it’s all new. Will I get more egress, when vetting has reached 100%? Before I upgrade anything, I want to understand, what’s keeping the bandwidth so low.
This is pretty normal your node is still pretty new and not very much data. Even my main node barely does 5Mbit up/down 10mbit total for a single node.
I remember when I used to get around 100Mbit to 200Mbit which was 3 years ago now and you will never hit those speeds with the amount of SNOS on the network today. At least not with a single node on a single IP you would need an army of nodes to really Max out internets, I have 3gig/up and down and I will never hit that even with 10 nodes running.
After it gets vetted by major satellites, ingress will likely increase about 5-fold, maybe more. Then egress will follow, as it depends more on the amount of data stored. Your numbers look ok for an unvetted node. Don’t worry and carry on!
Can this scenario be a practical “hack” for profits?
make a storagenode;
sign up for a free account with Storj DCS;
upload max data permited by free account;
quip redownloading that data as musch as free account permits.
Based on the premise that the closer the node with your data is from you, it will win most races and bring profits, this should be viable. I’m not familiar with the terms of free account, and I just make assumptions.
Yes, the data is distributed among many nodes, but some pieces may end up on your node too. So if the free account permits, you ca upload and download as much as is permited, with a script or something. Also you can make many free accounts with different credentials. Maybe not on the same IP as the node, but in the same /24 subnet, or at the same ISP.
Don’t get me wrong, I’m not interested in doing this, I’m too lazy, and don’t have programming skills. I just wonder if Storj is protected from such abuses.
Maybe others already doo this, I just want to rise the awerness.
Using the full free account with limits of 150gb storage and 150gb egress, taking into account the expansion factor and assuming you win every race your node was selected for, you’d have 31mb extra stored and 31mb extra ingress. Sounds like you’d be making bank… </sarcasm>
After the vetting has finished a one week ago, the ingress is way better and I’m getting around 1.7MB/s for the three nodes. This is quite good and I expect all three nodes to fill-up by the end of the year. This is cool!
