Boosters (especially for DC farmers)

#1

This thread is for moving the discussion around farming boosters forward, especially in response to farmers who have nodes in data centers but without gold certification. Here are my ideas for boosters to start:

  1. Reward public IPs for “capacity provided” in addition to the usage rewards. For some background, the reward model for public IPs was originally designed as usage based because we didn’t have a way to verify that they were actually valid or properly configured. The idea is that deployers simply wouldn’t pay for IPs that don’t work, so farmers would only get rewarded for those that do. I understand that the dev team is working on a way to verify public IPs and public node configs, meaning we could use that as input for this booster. The other short term possibility is that we somehow verify these in a more manual way and apply the boost at minting time without committing the data to TF Chain (not a very clean or decentralized solution)
  2. Bandwidth booster. Of course, some farmers at home are providing substantial bandwidth, so this alone isn’t enough to provide a real incentive for DC farming. We also don’t currently measure bandwidth at nodes, though this too is coming. So again, in the meantime we could use a manual method, with the same caveat as above
  3. Uptime booster. Simply looking at monthly uptime as a basis for a booster isn’t good enough, in my opinion. Many farmers at home achieve 100% in some or most months. The extra infrastructure in a DC should allow a farmer to maintain high uptime for months at a time. Also, DC farmers should have much more stable network connections, meaning that part of this booster could be based also on network availability (something that’s only considered in the current minting model in the case that nodes have a network outage that spans a minting period boundary)
  4. Utilization booster. This is the originally proposed booster and is perhaps the most important for aligning incentives on the Grid. It also works with the public IP concern, as nodes with public IPs available will naturally attract more utilization. This could start with a low threshold to activate the booster, which would then increase later as utilization grows on the Grid, for example to target the top 25% percentile of utilized nodes.

Of these, only 3 and 4 are available without any system changes or manual checks. They can be implemented simply through the minting code (potentially even retroactively for a number of periods if desired).

Do you have other ideas for boosters? For our DC farmers, what quantity of reward boosts would you need to see to feel it’s worth it to bring more public IPs or keep your nodes in the data center at all?

5 Likes
[Approved & Executed] The First Vote of the ThreeFold TF Chain (L1) DAO: Gold Certified Farming (CLOSED)
[Approved & Executed] The First Vote of the ThreeFold TF Chain (L1) DAO: Gold Certified Farming (CLOSED)
Incentive for farmers to attract workloads
#2

A booster for CPU’s over a certain benchmark could be awarded as well. Measuring that could be rolled into the same script for measuring bandwidth.

I think the incentives used to bring something to a DC should roughly offset the monthly expenses. I think a full rack runs about $1000. So an average TFT per rack would need to be calculated to get to the % boost needed. Probably something along the lines of 20%.

1 Like
#3

Not a DC farmer but a big NO for this, its a rabbit hole and generaly a bad message to do retroactive rewards IMO

1 Like
#4

Great inputs, thanks @FLnelson. CPU benchmarking is also in the pipe. I’d really prefer to work with on chain data, as that’s the idea for boosters, but I also think it would be a fun project to automate the deployment of a container to every node on the grid that runs these checks and phones home.

Generally speaking I agree, but with some exceptions. If a booster proposal passes and then takes longer than a minting cycle to implement, I think it would be okay to pay from the time of passage. Or in the case that a booster depends on multiple months of performance, I think it would be okay to start paying farmers early who met the requirement in the past (assuming this is on a rolling basis, not lump sums).

#5

If you need on chain metrics for boosters, then CPU performance would impact utilization. For example, Flux CPU’s benchmark requirements are very high and there is a lot of demand for that.

#6

Very good start, but I’d like to add another way to attack this. The current proposal is to add boosters for:

  • proof of public IP’s, not utilisation of public IP’s
  • proof of network (throughput)
  • proof of uptime (already exists, normal farming rewards)
  • proof of utilization (already exists, normal cultivation costs, not ending up in the farmers wallet

For me there should also be a number of boosters for having the following items in place and provisioned

Description Number Result
power per server 2x higher uptime
network per server 2x higher uptime
High throughput network >100Mbps better application / service performance
Multiple IPv4/6 addresses per server 5x allows for more direct connected services to the internet
Datacenter environment Tier 3/4 higher uptime, longer life for server

So - I also would like to investigate how we can reward the farmers that have chosen to take a proper datacenter environment with all the perks (power and network redundancy). I’ll continue to think about how we can “poo” this into a booster, an NFB, Non Fudgeble Booster… :slight_smile:

2 Likes
#7

My nodes are going to be in a D.C. eventually, it’s gonna say threefold on the door.

I’m about 80% done with the work on my first GEP which addresses this exact topic I plan to be done by mid day Tuesday us central time :). I think it will open a lot of doors for the discussion on this topic

#8

Good idea’s for all the boosters also for farmers at home. I just started a GEP to add Ipv4’s to home networks (still in the market) this booster will definetly help home users to get these IP’s, and i think on @weynandkuijpers opinion on “Proof” is maybe better dan utilisation since most nodes are not yet used on main net this won’t be going really fast with the many nodes there are. You got my word on trying with at least the dutch community to at least get 2/3 ip’s per 3node. Hope to do 5 really depends on the price they are giving :grin:. The a booster is a big PRE in the dutch community.

About ip’s in the DC i’m already Gold and have the option to get even more IP’s let’s say a /22 block 1024 IPv4’s and the option to upgrade to a full 10G fiber connection. With off course the suited hardware which is also not cheap so for that a booster will be suited.

I also agree on @FLnelson idea to boost the CPU’s benchmark and make it a little bit more intresting for people to buy valuable 3nodes and also make the people that already bought high CPU’s/benchmark happy.

And maybe with that differentiatie newer g9/g8 servers from g6/7 for example. It’s a different price class.

#9

Looking forward to this Weynand, glad you proposed it this way, makes good sense. We could even consider those to be Silver Farmers to give possible users another level of trust. Currently we could easily tick all those boxes, but we’re holding back on purchasing additional ip’s and connecting the second network as i assume this will require an additional switch looking at redundancy.

Speaking of which; it would be very helpful for both us as future users to have a video or HowTo on setting up Multiple 3nodes in a DC with two networks coming in, multiple Managed switches and a router, howto configure these properly and assign (and test) public ip’s.

#10

A CPU booster would help us selling certified nodes too :slight_smile:

Our CPU’s are brand new and super fast. These little quad cores are often better the Xeons in the servers people are running.

2 Likes
#11

Actually you forgot to add the SuperGreenFarming Booster, this one was already accepted but shouldn’t be forgotten. GEP for Super Green Farming with 100% Renewable Energy .

I don’t know what’s the plan on this GEP but it was accepted a 1.5 month ago. Maybe good to start a plan for that booster as well called “Prove of energy”

3 Likes
#12

@teisi thank you for the reminder. I have an update call with Jakub last Friday and we will update/conclude the discussion with regards to (super) green farming.

3 Likes
#13

Nelson, what CPU is running in the Peking and to what Xeon do you then compare?

#14

i3-10100 (4 core) vs E5-2670v2 (8 core) CPU benchmark 8,722 vs 11,541

The single core benchmark for the Peking CPU is much higher.

#15

I just want to ask a question, not trying to ignore the benefits of boosters. But everybody having a farm right now knew what TFT they were getting for their setup. Moving it to data centers, having more uptime, public IP addresses… it was all known upfront.

From reading all the comments, I do get the feeling that, now that the bear market has come to reality and the price of TFT is not what it was before, people are looking for ways to increase the amount of TFT with the setup they already have. I think we should be extremely careful in not just handing out TFT for existing setups. I mean, what are we going to do in case the price of TFT goes to 0.01cts? Increase more for the DC hosted farms so they can cover their costs?

Now on the boosters. Uptime is a great one and is a potential good one to overcome the problem that a node is switched off during the month as soon as the 95% can’t be met anymore (and thus, in future, zero rewards). It might make it zero for that node for the month, but knowing there is a booster of x-% if you have overall >95% for several months incentives it to keep it on.

I know we had a green-energy vote, but also there we need to be extremely careful in my opinion. Green energy does not do anything for the grid itself. ‘Green’ is in any case a very abused term and we should avoid centralization upon determining what is ‘green’ and what isn’t. For instance, my energy supplier has 100% renewable energy delivered to my house. Would that be declared as ‘booster eligable’ for green energy? Hell no (in my opinion)… we should stay far away from that.

#16

Ah ok, got it…I’m only running 16-core / 32-thread processors, all v3 or v4.

#17

The intent of the farmers for boosters is profit, but the intent of boosters for TF is increase the quality of the grid. We want to avoid any criticisms of running old hardware. We have not lowered the 8 cent starting price. That would have been disastrous.

You do make a solid point though. I believe we should keep boosters for certain DC and network farm improvements. But for new hardware being onboarded, lower the rewards for some of the current lower grade stuff being added.

For example, 1500 benchmark per thread earns at the usual rate, but between 1000 and 1500 earns 20% less. Lets be honest, those G8’s are not premium. This is not retroactive, just for new stuff.

I’m with you on green energy often being a joke. Somehow cutting down our local trees, pelletizing them, and shipping them overseas in bunker oil fueled ships to the EU is “green.”

2 Likes
#18

You are absolutely right that not all energy sold as “Green” is actually 100% renewable. It’s mostly sold as green due to the fact that we have a system based on “proof of origin”. That means you can potentially buy conventional or mixed electricity, where the provider guarantees that he will compensate some green energy producer (may be somewhere completely else geographically) for the purchased electricity. In the end some money that was paid will theoretically indirectly go to a green energy producer, but technically its’ more green washing than actually green, because it does not in any meaningful way incentivise more investment in renewables.

That’s why I keep advocating for applying potential boosters only to energy that is 100% renewable on a physical level. Meaning it has to be consumed where it is produced. Close to the wind turbines or solar parks and only during times of actual production (when it’s windy or sunny enough).

People and especially companies do care about their carbon emissions. Even if it’s just to look good in the eyes of the public. So having nodes powered by actual renewable energy can benefit the grid IMO.

#19

Well, the problem with ‘renewable on a physical level’ is that it is not green or good for the environment at all, and actually increases waste and CO2 levels. But that’s not really a popular (although logical if you think it through) statement to make.

Even if you have 100% wind energy, they only supply energy about 30% of the time. What nobody tells you is that wind energy is DC. In order to get it to AC, one needs a…powerplant. And powerplants are needed anyways to cover for the swings in energy provided to the grid. Those powerplants need to be able to turn on and off quickly, and therefore they have to be using gas turbines. Same issue with Solar. Nuclear is the only option for zero-emission energy, but it cannot coexist with wind or solar, as you cannot switch of a nuclear reactor that quickly. Plus, costs of electricity is huge because of the need of the backup plants being available - they are not running 100% of the time, so the capital is not fully used.

Then environmental issues with solar panels and even the blades of wind turbines. Solar panels are not reusable at all, as they can’t split the materials. In the US there are whole ‘farms’ of solar ‘waste’, just lying in the fields. Unprotected. Weather causes deterioration and leads to heavy metals getting in to the ground water. The blades from wind mills, same problem. They can’t be recycled.

Great example is Germany. They have the most ‘green’ energy in EU and there carbon levels have risen tremendously during the transition (and even worse due to their choice to shutdown perfectly fine nuclear power plants, which made them turn to brown coal again). I worked as an energy trader for a while and we had fun getting the German power AND getting paid for it.

Netherlands - same. We are suffering from solar panels, too many of them and the grid can’t handle it (the swings). We import wood from other countries (Baltics, Canada, US) as wood pellets, we burn them in our coal plants and call it ‘biomass’. While puffing out tremendous amounts of CO2. And call it ‘green’.

So… what you call ‘green’ or ‘renewable’ in my view is just not seeing the big picture. And no, I don’t believe it does anything for the network. Sure, there might be customers that would like to see their servers run on green energy. But you might as well then promote BLM, or any other political view as there will be people supporting it.

Bit of long story and controversial :slight_smile:. So I"m not in favor of having ‘green’ energy being rewarded extra TFT (too late, I know as the DAO ruled otherwise). Also, how do you prove it? it’s too open for abuse, depends on someone certifying it. Not open, not transparent, not decentralized.

1 Like
#20

Sorry but what you are talking about here may be true for some edge cases (if implemented completely wrong), but this is not true as a general statement at all.

Our wind turbines produce between 80%-90% of the time. When combined with solar we can reach up to 95% availability. Not sure where you get your numbers from, but we actually own wind turbines and have all the data for those machines.

True for the present. This will change in the future. Our energy grids will have to become much more flexible. Energy storage capacity and flexible consumers (possibly even 3Nodes) will make it possible. You have to see the long term developments here.

Obviously there has to be made an effort to recycle everything properly. If it’s not economically feasible to recycle, then proper incentives have to be introduced to built up the needed infrastructure and capacity. It’s wrong to say that it cannot be done. It’s possible and much cleaner and safer for our planet than currently known alternatives.

This is just not true. The opposite is the case. Especially in the energy sector. Seeing that you’re claiming this makes me wonder where you get your information from. You can verify this easily by looking at publicly available information.

If that’s what you are doing in the Netherlands then I hope you’ll reconsider and increase your solar and wind power. That’s the way to go for most countries.

If you think the transformation of the energy sector to become more sustainable has has anything to do with political views I don’t think you have understood how seriously we need to reduce green house emissions in order to keep our planet as beautiful and habitable as it currently is. It’s not about politics, for many people, especially in the developing world, it’s about survival. Don’t believe me? Thats completely fine. I am not here to persuade you. But I do feel the urge to clarify this in case others are following this thread.

Everyone participating as a green farmer will have to get certified by ThreeFold one way or the other.
I can’t speak for others that may join the grid as green farmers in the future. Maybe there will be people trying to cheat or use “green washed” types of energy. I can only tell you that the way we’re planning to participate in this will be one of the most sustainable and environmentally friendly ways. By using wind and solar power from our own production. Right where it is produced, right when it is produced.