First of all, to clarify - running a witness node, and running a "full" RPC node are two different things. Often the witnesses are the ones who run full RPC nodes, but technically speaking they are not the same. (Not all witnesses run full nodes, and not all full nodes are run by the witnesses.)
In terms of the system requirements for running nodes (both witness and full RPC) the development team provided a lot of information about this back in this post.
If SMTs lead to more adoption and more transactions on the blockchain, then that will ultimately lead to higher requirements for running nodes. The hope is that greater adoption will lead to increase in price, but it is not a guarantee. That is always going to be a risk for the economics of running a node.
The cost of running a witness node is still relatively low, and even with some extreme growth - it should still remain reasonable for top witnesses to afford running a witness node for the foreseeable future, assuming we don't see an extremely drastic reduction in price.
As far as full RCP nodes, those could start to get a lot more expensive if the memory requirements exceed 512 GB of RAM, but from what I've heard the development team is planning on expanding the RocksDB implementation (that was recently adopted for account history) to support the other operations as well - which should significantly help with the memory requirements for those nodes.
Steemit Inc / @steemitblog have claimed that full RPC nodes require 32 GB to 64 GB of RAM post RocksDB/AppBase/0.19.1x, and that the recommended configuration is shared-memory on NVMe/Optane disks even for full nodes. RocksDB is repeatedly heralded as "on-disk" solution. (Source; and Source)
I'm not aware of this being verified independently, however. Either way, for 0.19.10+ the "512 GB RAM" requirements are questionable, at best; obsolete most likely. I have no reason to believe Steemit Inc. are lying.
In short, the cost of running a witness node is very low and you get paid. The cost of running a full node is insanely high and you don't get paid.
A full node costs $700-$1,000/month, and soon to be much much more. It's not cheap, yet a witness node costs $60 and pays anywhere from $0 to $100K+/year. Not saying top 20 shouldn't run one, but it shows the flaw in the system.
A full node has nothing to do with witnessing, and this is coming from some who manages 4 full nodes. Full nodes are a dev project, and a project the size of DTube shouldn't be using public nodes and should have their own, just like any large projects, especially with 2M free delegation.
Public nodes serve one purpose, they allow small projects to bootstrap and avoid the crippling expense of a full node allowing them to bootstrap. Once you get large enough, you should be using your own private full nodes to sustain your project not leaching off the very few public full nodes that are available and overwhelmed.
Hi, why full node cost so much? what bandwidth/month please?
Because of the ram requirements (512GB+)
oh, and mayby that can be put on large ssd to do the ram work?
because im interested to host such from home, no way i can have so much ram.
Using ssd would require 4 week or longer to replay.
I agree. This needs to be addressed.
Hey Tim,
Should you and other top witnesses not setting price feed bias percentages, with a SBD Debt Ratio at 6+% ... there is to much steem being printed.
Or is it that you and the witnesses know this an play dumb? Because at 10% the SBD floor will be gone, and a bail-in is just what you guys want so that users and community be pickpocketed?
Who else is going to pay to bring that 15+ million SBD Debt down? Who is going to burn it?
Looking forward to your answer.
This is extremely off topic for this thread, but I will reply non-the-less.
I am not planning to use a price feed bias as this time. I am not playing dumb. I am aware of the economics behind SBD and have written quite a few posts about the topic. I also submitted two pull requests to update the SBD economics as part of hardfork 20.
There is little need right now to "burn the debt". The market has demonstrated over the past several months that there has actually been a demand for more SBD. A short-term change in the cryptocurrency markets is not enough to warrant any drastic policy changes.
Even if the debt limit exceeds 10%, that is not the end of the world. The price of SBD has been sustained for a prolonged period without traders even using the conversion function. If conversions are not supported at
1 SBD to $1 worth of STEEMfor some time - that does not necessarily mean the market will not be able support the peg based on demand alone.I'm not really sure what you are referring to as a bail-out. I have heard no such thing being discussed, and it doesn't sound like something I would support.
In terms of the absolute worst case scenario of what could happen, the debt ratio could significantly go past the 10% mark, and the market may see that as too substantial of a risk, and no longer support the peg. At that point the SBD peg would become broken (well, more broken than it already has been, given that it has not really been at a stable $1 USD price for quite some time). The risk here is 100% on SBD holders, and poses little risk for holders of STEEM. This is really more of a "hypothetical what-if" though, and not something I believe has a high probability of happening.
Tim, thank you for your reply,
The bear market has significant downward pressure on steem and also sbd.
Because of this, the market value has dropped also significantly.
This is a new situation because, with 15+ million sbd debt and the market cap in a downward spiral, they will meet each other at 10%. Because we stopped printing sbd's at 5% there is more steem created. Which re-enforce the downward pressure on steem. What we see is that even when steem price is rising, the market cap is still coming down. Price Feed Bias is created as control mechanism to keep stability.
Your proposal to extend the debt ratio from 2-5% to 9-10% is like opening the door for a "Black Swan" event to happen. Before the HF20 we could already be above 10%. Will you put HF20 on hold?
Past performance is no guarantee of future results.
At 10%+ we break the SBD floor, this means sbd will free float and could go to zero. And your mention that only the sbd bagholders will feel the pain? Is just not true, in a scenario like this, steem would go to zero first, followed by the sbd bagholders. Remember, SBD is not being printed, but steem is, more and more. It's like holding a little bird in your fist, and sqeeze all the life out of it.
Then there is only one more thing to do, and that's close the door on our way out.
Based on some of your comments, I don’t know if you fully understand how SBD works.. I’ll try to do a post in the next few days to cover the topic and hopefully your concerns.
It's a problem but I don't see it going to zero. I see it going back to the pennies it was last year which is still pretty bad.
Agreed... but...it can be really really bad, for everyone...
Here is a more useful question. Will interest again be applied to SBD so that people have a reason to hold that? Is there anything which can be done to drive demand for SBD and Steem?
The downward pressure is is a real problem. It makes life harder for everyone including witnesses.
I have no immediate plans to add interest for SBD. I don’t think it is necessary under current conditions.
Generating demand for STEEM and generating demand for SBD are largely unrelated.
As far as increasing demand for STEEM, my most recent post is touching on the subject. The short answer though is that it is everyone’s responsibility.
An interesting question to think about, because many stakeholders think this way - if someone’s content is not generating more demand for STEEM, should it even be getting paid?
You give great comments based on knowledge and experience. Thanks!
Glad you came in to clear a few things up, Tim. Always good to see you coming in with the wisdom.
@timcliff I appreciate the response, however I think the question remains. How would the decision by the Witnesses not run the full RPC node impact the performance of the STEEM blockchain once SMTs go live? Will we begin to see transaction times closer to Ethereum and Bitcoin or will it be negligable?
The transaction times are three seconds, and they will remain so even with significant growth.
@timcliff thanks for the response. Honestly, as a non-technical person I have no idea how any of this impacts the performance of the blockchain and I appreciate the insights you’re providing.
So are you saying steem is getting in over its head with SMTs?
I don't know how you got that from what I said. No, that is not my assessment.
Quoting one of my favorite videos I watched recently by @dhenz: "when prices go low, people start asking weird questions."