You are viewing a single comment's thread from:

RE: PsyberX RichList

in 1UP3 years ago

The produced blocks depend a lot on how high you are in the queue (aka how many people are voting for you), so it's not a static thing. But if you wanna have a rough estimate "starting" like my @atexoras.witness that is running a node since the beginning of the year, so a month. And from all the signed blocks, so far, I have received around 15 HIVE. Hive from blocks signed go directly to powered up HIVE.

But, in my view, I would not be thinking of running a node for profit. To be honest, the real reason for me, was because I needed more decentralized access to HIVE (reaching my node instead of reaching a node 20 000 KM from me, makes a huge difference). Hence add another node to the ecosystem, that currently only serves me, but if things evolve into the right path, I will then be exposing my node, so others can benefit too. For example, people in New Zealand, Australia, and all islands around. Which will benefit from shorter interaction with my node, as compared with others further away. This is the natural path of decentralization... we get bigger and more options and fewer single points of failure.

Resources-wise, if you have a very good IO subsystem (very low latency, live NVMe's) you can run everything from storage, hence you can easily start a witness node with 2 or 4 cores and 4GB of RAM. Even less most likely but you should understand that replaying the blockchain will consume 1 thread for the entire process plus something else for IO depending on your storage configuration/type.

If you deploy what is called the shared memory file into memory (using a ramfs filesystem), then you will need around 20-21GB of ram JUST for this file alone. So, I would say, at least 24GB VM or System. CPU wise here, you will want the highest clock CPUs possible because all you want is reduced latency. 4 might be a good core count, but you won't be using lots of it most of the time. It depends on who will be using the node. If just for signing blocks, it will use almost no CPU. HIVE code is super efficient at the moment.

Finally, I would definitely recommend starting with things on a Virtual Machine (cheap one) if you wanna just give it a try and then move to a physical one. But if you already have a physical one doing nothing, then you can use that too. Physical vs Virtual will always be best, because of the extra latency (even if not much) the hypervisor introduces. But, for most cases (like just signing blocks) it will be negligible.

Have fun! 😎 HIVE rocks!

Sort:  

Is there a page that shows someone like me how to get started? I have a desktop machine that I don't use for much. It is an older machine that I would likely need to upgrade, but that's something I've been considering anyway. I probably wouldn't go less than 32GB of Ram, just because I want a clean experience on my system and it would likely be a i7 at worst, so the demands don't sound like they'd weigh my machine down. Does running a node weigh down the internet connection much? My family does enjoy our Netflix!

Searching through posts (use ecency search bar for example), you can find lots of good ones. Some examples I grabbed from a post of mine:

Does running a node weigh down the internet connection much? My family does enjoy our Netflix!

And no, it should consume very little some bandwidth, but nothing that a +20Mbit connection can't deal with (check my other comment). I don't have number right now, but for a witness node is very little. With exception of the initial download of the blockchain which is a little bit more than 500 GB.

You're awesome...looks like I'll get a separate hard drive for this, too...I've got some homework to do...

There you go...

image.png

A week worth of traffic (includes a hungry Hive-Engine node feeding on it):
Upload 7.23 GB
Download 29.00 GB
Total 36.22 GB

Thank you so much!