Ingress and Egress of data

HI people. I am an SNO that currently runs storage nodes in 4 different locations. since storj doesn’t work like most common services latency is a relative term to target. I am wondering and plan to find out down the line this.

3 of my 4 nodes are new and still getting vetted. Non the less 2 of them are located next to one of the countries IX`s (Internet Exchange) I currently do not pay much hence I am able to operate SN in a DC. but since they are next to IX they have super low latency to surrounding countries in Europe. Meaning that relative latency to lots of places in Europe is low as the IX is liked to others in different countries using a high speed link.in other words people that want to utilize the node theoretically have smaller latency.

Now the question is are nodes with lower latency more likely to be utilized?
And since he latency and upload speed is higher will they have a chance of having a higher Egress?

If anyone has any experience let me know. Thaks

1 Like

Hello :slight_smile:

Interesting question, from the point of view of the uplink client, latency isn’t one of the current variables used to asses a node selection. When the client wants to uploads data, the satellite sends the list of random IP’s, based on subnet, and anything else storj adjust in their database (we don’t know how the data is stored, and not saying anything bad just saying we don’t know [not storj bashing]) - back to the uplink client. Not all the nodes need to respond, to have a successful upload, so if your node has low latency, and very quick disks, and responsive software you stand a good chance of beating other slower nodes to upload, and storing the piece ~ 4mb max.

You also need to take into account, that client latency to you will be an impact - geo location doesn’t matter - if you in europe, and client in europe the assumption that latency to you is better than client to americas is incorrect - the clients ISP could have a better L1 / L2 peering to the US and have less latency then L3 local DC.

I think that having a low latency, very quick disk node will help you keep a good online score, and for the traffic sent to you, should mean that you have a very low cancel or fail score, but it from my understanding will not drive more traffic to your node - this is a good thing :slight_smile: decentralized storage, that priority low latency nodes would very quickly not be decentralized.

The interesting thing, and something I haven’t looked at yet is Ingress Repair traffic - this does seem to favour longer established quicker nodes - but this could be due to slower, unstable nodes are more likely to be still in vetting, and don’t get full load - think only 5%

Egress, I haven’t looked at yet - I’m assuming it’s still the same that the quickest nodes get the credit.

I haven’t found quick way of getting ingress data, run a quick stable always online node seems to be best we can do :slight_smile:

I personally wouldn’t spend lots of $$$ on expensive hosting, until such time that you have a good few TB of data, and are at lest 6 -12 months online - the held back amount is far to high initially in my view, and you risk loosing it if your node crash - depends if you like gambling :slight_smile:

CP

2 Likes

I’m quite lucky as I have to run the server non the less as there are some other services that require the stability of DC. Just part of the two servers is given to the storage node. So the overall cost is technically low for me. And the over all servers have a relatively cheap hosting fee. So if the worst comes I can alone cover the cost of running them, they would pay for them selves. I estimate that will happen at 7month mark.

Thanks for the quick response.