[Witness evaluation post #1] These witnesses run a full node. Voting suggestions if you care about decentralising the blockchain.

in busy •  10 months ago  (edited)


I'm busy doing some research tonight for my personal witness evaluations. I'll share them here. This is not final, things can still change. My evaluation posts will probably grow a lot in the future. If you want to let me know what you do as a witness, you send it to @exyle on discord or send a mail to [email protected]

Any mistake in my research I will correct.

When you evaluate a witness the easiest metric to use first is looking if they run a full node.

It's a binary question. Either yes or no. There are no emotions involved.

Steem inc. has money issues and they have let us know that they need to reduce costs. (They are working on reducing the RAM requirement of nodes reducing the rent costs, but it's not done yet).

They currently run full nodes that app developers that can't afford a full node use to make their apps work well.

Running these full nodes is expensive (an estimate I heard is around $10k a year each, could be more, still trying to hear the right assessments).

But we all need them. Without full nodes, no Steem blockchain. Simple.

To decentralise Steem and take the pressure of Steem inc. it's therefor very commendable when witnesses, outside of Steem inc., run a full node themselves.

They can either run a full node private or public.

Public nodes benefit other developers too which is nice but private nodes are also nice because you don't put strain on the public nodes when you run your own infrastructure.

There is no perfect documentation about who runs what so, I used @fullnodeupdate (made by @holger80) and a post made by @timcliff here to guide me.

The following witnesses run a full public node that I found out so far. This list can change in the future.

@gtg (top 20)
@timcliff (top 20)
@followbtcnews (top 20)/ @aggroed (top 20) (run one together with minnowsupportproject)
@ausbitbank (top 20)
@anyx (top 20)

The following witnesses run a full private node:

@good-karma (E-steem infrastructure). (top 20)
@thecryptodrive (top 20) / @reggaemuffin (@buildteam).

From https://steemd.com/witnesses:

Top-20 witness pay is $3,000 per month (approximately) based on the current feed price.

If you as a Steemian value the witnesses that run a full node and care about the decentralisation of the blockchain then these witnesses are voting suggestions.

I am part of witness @blockbrothers.

Please consider us for your witness vote if you think we deserve it here:

Vote for @blockbrothers via SteemConnect
Set blockbrothers as your proxy via SteemConnect

We are the creators of Steemify a notification app for your Steemit account for iOS.

Get it Here:

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

I wrote in my most recent witness update how, in my opinion, it's not necessary for the witnesses to do all of these things themselves, but just to make sure that they get done. I've considered running a full node, but I realized that the only thing that would really do is look good and get me on these lists.

Just throwing up a full node on a single server isn't sufficient or reliable enough to support any serious projects. I have been pushing for a long time for there to be an enterprise-grade, private, paid Steem node service, but that's prohibitively expensive right now because the tooling isn't there to support it.

$2M / year is an unbelievably high amount for steemit inc to need to spend to run their node. Projects like Jussi and RocksDB were supposed to help bring that down, but from what I've experienced and heard from others Jussi is an absolute mess and after 6+ months of work the requirements for running a full node don't seem to have been reduced at all.

I think it's good that this market downturn has brought all this to a point where it can't be ignored anymore. As a witness, it's my job to make sure that the proper tooling is built to severely reduce the requirements for running Steem nodes and then then make sure that adequate nodes are available to support all projects that want to run on the platform.

I've recently been working with @anyx who is doing some great work towards those goals and we are working on switching steem monsters to use his nodes instead of steemit's in the very near future. I will also do what I can to support him and anyone else who is working to make things better, which I think is a better use of my time and resources than just paying to run yet another full node which can't actually support any serious amount of traffic.

Thanks for taking the time to write such an extensive comment. I have learned a lot from it.

Would you say that when RocksDB and Jussie don't work out, Steem is looking at a serious problem if the bear market stays for a while longer? (cost wise) or can it be solved in another way?

We're already at the "serious problem" point, but luckily there's an amazingly talented group of people here who will solve it, and are already making great progress in an extremely short amount of time. It may be a bit bumpy over the next couple months as changes are being made (such as the recent node updates announced by steemit) but we'll come out much, much stronger, and more decentralized, on the other side.

I've considered running a full node, but I realized that the only thing that would really do is look good and get me on these lists.

Just throwing up a full node on a single server isn't sufficient or reliable enough to support any serious projects. I have been pushing for a long time for there to be an enterprise-grade, private, paid Steem node service, but that's prohibitively expensive right now because the tooling isn't there to support it.

^ This.

I've been thinking about a public RPC node as well a few months ago, but decided against it due to

1.) extreme costs for someone outside of TOP 20 (already running 3 nodes, which are exceeding my budget) with no way of knowing if RAM usage would exceed 512 GB. It seems that due to the bear market, HF20 and lower usage we are not hitting that target yet, but still.
2.) it's much better to have private nodes for projects / SaaS where users can pay for them, than offering public ones. They will be hammered away and become unstable / unreliable. (countering this requires a lot more power, which is why Steemit Inc. is paying so much for server costs)

Now, I'm not saying that API nodes aren't important, but I personally would wait until RocksDB is implemented fully (which tremendously reduces the required RAM) before I'll spin one up. And I actually find the idea of SaaS (meaning, projects can pay for requests to the blockchain/have a limited amount) a lot more interesting.

In my opinion, it's important that we're not only reacting to events (for example HF20: many Steemians cried out about witnesses doing better code-review, are able to code on a blockchain level) and now that suddenly RPCs nodes are in the focus, witnesses are judged if they run a RPC node.

Instead, I believe in focusing on longterm solutions. For now, the network is completely secure and I'd rather see work on longterm solutions, instead of short-term reactions: witnesses spinning up RPC nodes. (see the point 1 & 2 above for my reasoning)

But anyway, I appreciate your post and this initiative, @exyle!

  ·  10 months ago (edited)

Thanks for chiming in, man. I like the idea of Saas. I'm assuming that results in if you use it more you pay more.

Of course! If every TOP 20 would run an API node after rocksdb is implemented, that would be good enough in my opinion. And projects could either use a private node themselves or even share a node between projects.

I like the idea of Saas. I'm assuming that results in if you use it more you pay more.

@upheaver actually did some brainstorming in that area, so he should be able to talk more about it, but in a nutshell: developers/projects would pay for their usage That way, they wouldn't need to either host heir own or share it with other projects.

That's right, most of the apps on steem are currently too small to support running their own node, but they would have no problem of paying for a usage based API access which scales together with their business.

(I couldn't leave that without comment, but I don't have anything to add and single word looks spammy ;-) )

I totally agree with everything you say within the brackets (oh, and exactly!)

lol, well there's a lot more to a single word comment than just the word itself. For example, an "Exactly!" from Gandalf means a lot more than from someone else :-)

  ·  10 months ago (edited)

It's ok, man. Glad you showed up. If I overvalued the full node argument than that's what is. I have no problem shifting if the facts change. Just grateful to learn more.

Top 20 pay = sustainable to run a full node at current pricing of STEEM. I'm pretty glad that I dont need to change many of my existing witness votes to be supporting those that run full nodes.

I analyzed the full node performance over the last 30 days here:

The full node from @bitcoiner is offline.

  ·  10 months ago (edited)

This is why I love to use the community (by making a post) to help me sometimes instead of going at it alone. Thank you so much for the work you have done there.

I used to run the two full nodes that BuildTeam sponsored, but the public nodes are no longer available.

  ·  10 months ago (edited)

Thanks for letting me know. Just to be clear. You run no public node anymore? (I'll take you of the list).

Not currently, been asked by a few to run some private ones for them as the ones I managed were very fast. I have considered many times to run a full node but a public node is a nightmare. Without a good way to limit the type of queries people do and the lack of proper visibility of what is being done, it's a big PITA and for the most part, no one cared if you ran one or not Steemit Inc node's might go offline.

seed.bitcoiner.me I don't think is a real full node, it never worked for me.

I do still run a node, just not public.

They were the fastest API nodes on Steem. True story.

I think that voting for witness just because they run full node is not good idea.

There are many low rank witnesses that can't afford to run full node.

It should be a must for ALL top 20 to run a full node. It's not that expensive.

Posted using Partiko Android

The full node is important either public or private.
Let's push for more to alleviate #steemit burden.
Since Steem is decentralized I wish the witnesses that run a full node if they need help, request it from the steemians that vote for them through support and delegations.
Let's make it a business decision if it deemed to be, so support can flow to the right channel to safeguard #steem blockchain.
Keep on postin

Exyle.nl interesting. Are you going to create your own Blockchain?

Curie is also top 20 :)

It seems like a good trend would be to see more witneeses move towards this as it will improve the decentralization. Is this something that a person who is not yet a witness could consider to assist?

Posted using Partiko iOS

Yes, anyone could run a full node. It doesn't really have anything to do with being a witness. Steemit, Inc. for instance does not run any witness(es) but they run the default node for steemit.com and many other DApps (api.steemit.com)

Thanks @exyle for showing some transparency on the block producers.

Posted using Partiko iOS

thank you for doing this research. i find it really valuable as we assess how to stabilize a more decentralized ecosystem!

I love this Kind of posts exyle!
I upvoted with 100%.
I love to see your points and the arguments in the comment.
Loving it and waiting for more my friend!

I worry what is going to happen when steem goes to 10 cents and the witnesses decide its not worth it anymore

Better ones will take their place.

Thank you for the suggestions ^^

The issue is nobody knows the focus of steemit inc, including steemit inc itself. Without clear decisions being made rather than keeping options on the table talk, I've lost a lot of confidence.

Time to do some voting. Thank you!

Thank you for your work ad post. When I began with the witness servers I wanted to run a full node also but it needs to be paid because running a full node with only 256gb memory costs like 200 euro a month. So first I was held back by this and thought first see if this pays the bill my servers costs are like 50 euro a month. on the backupserver I run a rpc its buzzing like a cittycat right now. I'm happy with the new solution coming up as soon as they impement it so that I can run it on a 60gb server then I will rent one. Or maybe if it is so small I will run it on my backupwitness.

With your document in hand I wil change my witness settings too.

Great post !

I am a neophyte in this and these doubts arise:

@exyle what is necessary to have an active Full node?

why do you say this?:
"Top-20 witness pay is $3,000 per month (approximately) based on the current feed price" because of the necessary machines?

What is the benefit of having a full node?

thanks in advance!

firma 2.png

Very cool! Thanks for this list. I will have to dig back into my witness votes and do some adjusting in the coming week.

Reading the comments on this post is as interesting as the post itself. [Looks up nodes]

Thanks for this info @exyle.

Just checking, is @privex the same as @someguy123 ?

Many thanks