Just as the rise of DeFi tokens has been rising wave after wave, another hot spot this year, Filecoin, is quietly advancing. On the morning of August 25th, the incentive test network (or space race, big miner race) was finally launched, attracting hundreds of miners to participate.
As of 21:00 on September 2, Space Race has mined block 24822, the number of active miners has reached 378, and the storage capacity is 84PB. During this process, network bugs and mechanism problems were found out one by one by the officials and the miners. At the same time, the technical solutions of the big miners were also tested in actual combat. Therefore, the results of the competition have attracted much attention.
for accurate understandingMiner ranking metrics, implications, and current and future goals and challenges for the network, Odaily invitedLi Bai, Technical Director of Pomegranate PoolBe a guest in the Chaohua community for in-depth analysis.
According to official data, the current effective storage capacity of Pomegranate Mining Pool is 7.4PB, ranking first in the world.
In the dialogue, Li Bai briefly introducedThe mining scheme adopted by the pomegranate mining pool - that is, to choose to configure more than 20 small and medium clusters instead of several large clusters. There are several clear advantages to this approach, he says.
"This arrangement can maximize the efficiency of computing machines without wasting performance. It can also prevent single-point failures and have a better fault tolerance rate. We will not say that because a few machines are down, any bugs will affect the efficiency of the entire cluster; Even if something goes wrong, it’s easier and less expensive to reset a cluster.”
Li Bai then analyzed the ranking criteria of miners.
"The data dimension selected by the space race is actually very single. In addition to computing power, which is the success rate of receiving orders, there are actually many important data in the Filecoin network, including the number of blocks produced, computing power growth rate, etc. We can't just look at the list Looking at the screenshots, it still depends on the continuous performance of the node for a period of time, stability is also very important, and the combination of several factors can comprehensively measure the strength of the node.”
For the current pre-stake rate of about 0.57FIL/Sector and the 20-day follow-up mortgage, Li Bai calculated and inferred that miners on the entire network may have to mortgage more than US$100 million in coins to mine normally. With so much money, I believe the official will not do this.
“The economic model will definitely be revised, looking at the latest document, the official plans to allow post-mortgage block rewards to be added to pre-mortgage, which is a positive signal. Otherwise, in the case of low liquidity in the early stage, miners' hard demand for mortgages will have a great impact on the market. "
secondary title
The following is the AMA essence organized by Odaily:
OdailyQ1: Manufacturers have different ways of making the list. I think your characteristic seems to be scattered nodes, which brings visual impact in the form of the whole screen (give us a picture to feel for yourself)? I would like to ask why you adopted this strategy, what is the current overall data situation, and how much can it be ranked?
Li Bai:Hello everyone, I am Li Bai from the pomegranate mining pool.
Our participation in the space race this time is indeed different from the style of the testnet, so a few days ago, many friends from the community came to ask where our nodes went.
We did not use large clusters to deploy this time, but used small and medium clusters that are more in line with the production environment, with a total of more than 20 deployments.
Screen swiping is just a side effect haha, these nodes are basically distributed between 20-50 as in the screenshot. This is mainly due to considerations of efficiency, security, scalability, and so on. For us, such an arrangement can maximize the efficiency of computing machines without wasting performance, and it can also prevent single points of failure and have a better fault tolerance rate. We will not say that because a few machines are down, some bugs will affect the efficiency of the entire cluster; even if a relatively large problem occurs, it is easier to reset a cluster and the loss is smaller. On the other hand, the hardware configuration and so on between our different clusters may not be exactly the same. Using a small cluster will facilitate our operation and maintenance.
We have specially built a system for operation and maintenance and early warning, so that more clusters can be easily added in the future, and the scalability is better. For example, we launched two new solutions, software service and hardware cluster a few days ago, that is to say, if you have hardware, but the mining efficiency is not good, you can also only use our software service, which can conveniently use our current technology and successful experience. Copy it out to help more miners improve mining efficiency. For us, this is equivalent to expanding one or several clusters, which is very convenient for monitoring, operation and maintenance, and deployment.
In terms of data, the total computing power of all our clusters is above 10P, and you can go tofilfox.infoGo up and see. Since filecoin’s official miner registration can only include 16 mining nodes, that is, only 16 clusters can be placed under one account, so our official ranking is only 7.4PB, currently ranking first in the global list, 1.8 more than the second PB.
OdailyQ2: The testnet has been postponed several times and finally launched. Were you excited at the beginning? Can you tell us about some interesting mining experiences during this period?
Li Bai:This rewarding website said at the beginning that "it is likely to start in early July", and then the actual start time is at the end of August. It is a lie to say that it is not exciting.
However, the official start time was selected at 6:00 a.m. Beijing time, which greatly affected sleep. Many of us got up at 3:00 a.m. that day, waiting to deploy nodes, receive test coins, etc. after going online. As a result, I did not expect that the new network encountered a bug at the height of 101, and the block generation was stuck.
Then there were many other bugs in the first version, and then the official released 4 versions a day, and they were constantly updated. Up to now, 7 versions should have been released in a week, and there are still many things to do in various follow-up updates and maintenance.
Then there are network security issues and order acceptance issues. As you all know, there were many situations in the first few days when it was launched, so it has been in a state of constant rotation recently, constantly fixing bugs and coping with new situations.
OdailyQ3: I see that your browser is quite easy to use. It not only lists the entities corresponding to many node codes, but also has a lot of characteristic data. Can you introduce it to us in detail?
Li Bai:Wefilfox.infoWe still put in a lot of effort, and the official came here to praise us.
At this stage, what everyone is most concerned about is of course the space race and the miner rankings. However, the data dimension selected by the space race is actually very single. In addition to computing power, which is the success rate of receiving orders, there are actually many important data in the Filecoin network.
For example, in addition to "effective computing power", our leaderboard also has two ranking methods: "number of blocks produced" and "computing power growth rate", which show the number of blocks produced by miners and the change of computing power in a short period of time. The logic behind this is that "computing power does not equal block production does not equal income". With the computing power, you need to produce blocks to get rewards, so we will have a parameter called "lucky value" in the "number of blocks" ranking list, which is the ratio of the actual number of blocks to the theoretical number of blocks, you can check Miners' optimization of bursting blocks. Then there is a parameter in the "computing power increase" called "mining machine equivalent". Based on the official configuration and computing power increase time, we estimated that some miners are equivalent to several "official mining machines", which can be used to estimate The size of the miner. They are sorted according to "effective computing power", "number of blocks" and "computing power growth rate". Of course, these data are just some references, and the architecture and optimization degree of each company are different, so it is necessary to look at it comprehensively. You are welcome to explore more functions. Recently, we will also publish a special page on the space race, which is better than the official one. Haha, you can go and have a look.
OdailyQ4: Many node vendors will make rankings and promote their achievements to investors. But these data are instantaneous. Looking at it over a long period of time, what data should we use to measure the strength of node vendors?
Li Bai:I think that when looking at the list, you can’t just look at the screenshots, but also look at the continuous performance of the nodes for a period of time. Stability is also very important. "Stability" can be measured by looking at whether the computing power of miners is growing steadily, whether there is any loss of computing power or coins being deducted, plus the "block production ability" mentioned above, several factors can be combined to measure the node more comprehensively Strength.
OdailyQ5: After only two days of excavation, nodes such as Diancun, Shikongyun, and 1475 encountered DDOS attacks, and the storage computing power of multiple nodes was even deducted all at once. The theory of "Song Jiang's attack" has intensified. Can you tell everyone the cause and effect of this incident? Does "Song Jiang" really exist?
Li Bai:"Song Jiang" cannot be faked, after all, slack and medium accounts need someone to write. But I doubt whether he will attack the network, because the attack requires technology, money, and time, and an attack without any benefits is not worthwhile. And another point is that at the beginning of the network attack, the loss to the miners is not large. Anyway, there is not much computing power accumulated, and it is a big deal to start again. After everyone is prepared and takes protective measures, it will be very easy to attack again. Difficult.
As for the DDOS attack, because in order to participate in the space race, everyone has to expose their IP, so there will definitely be some attacks, which is also part of the daily work of the mining pool. Large mining pools and exchanges should be subject to a large number of attacks every day, so they must be well protected. As for who is attacking, only the attacker knows, so it is not easy to guess casually.
OdailyQ6: Before the mainnet goes live, besides participating in the competition, what preparations should miners make, such as raising coins for pre-mortgage, etc. How will this affect the secondary market?
Li Bai:The problem of pre-mortgage is still very serious. Although the official faucet is now issued on a daily basis, there are still many miners who cannot get enough coins to mortgage, so they have to go to slack to "beg" online.
Now the official has lowered the pre-mortgage standard, and each block can mortgage up to 1FIL, which is much friendlier than the previous design. But this still requires a huge amount of coins, especially the post-mortgage makes the block rewards have to wait 20 days to start a 180-day linear release, so that the block rewards cannot participate in circulation and mortgages in the first 20 days, which puts a lot of pressure on miners big.
If it is calculated by adding 10PB of computing power every day, 310,000 FIL will be mortgaged a day, which is more than 6 million U.S. dollars at a unit price of 20 U.S. dollars. If it lasts for 20 days, miners on the entire network will have to mortgage more than 100 million U.S. dollars in coins , miners will not have so much money.
So I think the economic model will definitely be revised. Looking at the latest documents, the official plans to allow post-mortgage block rewards to be added to pre-mortgage. This is a positive signal. Otherwise, in the case of low liquidity in the early stage, miners' hard demand for mortgages will have a great impact on the market.
As for how the price will change, it depends on how the economic model of the final mainnet is designed. I believe that the project party does not want the short-term price to go up to the sky, and everyone cannot mine, so it will be meaningless.
OdailyQ7: There are a lot of controversies on the Filecoin project side, from repeated delays to the undecided pre-mortgage design, resulting in endless voices of forks and "attacks". Do you think the official series of actions are justified? Have you seriously considered the feedback from the community (seeing an anomaly that the mainnet launch time may be advanced)?
Li Bai:There is still quite a lot of communication between the filecoin project party and the miners, at least we have frequent communication with the official, and various bugs and economic model problems have also been reported in the past. Judging from the recent changes, the project party is still listening to the voices of miners. I understand that they also need to experiment, to build the final model.
The relationship between the official Filecoin and the big miners is actually a relationship of mutual cooperation but mutual game. How to design this mechanism to generate enough incentives without affecting the long-term development of the network due to the behavior of the miners is a very difficult problem. problem, so even more people from both sides participated, and it is normal to discuss and revise.
Blockchain is an open source, open, permissionless network that is also a community. The official has always had a very close relationship with the miner community, and the official can also see that the coins in the faucet are not enough, so it must be insufficient after the main network. I believe that these problems can be solved on the main network.
OdailyQ8: It stands to reason that the miners should be the most affected by the "pre-mortgage", followed by the miners. The miners at least earn money from selling machines. Of course, the miner's repayment will also affect his share, but in the end it is still Miners are greatly affected.
1) Then why are most miners opposing it?
2) Will miners have too much say in the entire mining industry, or even in the entire community? In this case, there is a risk of splitting the community. The emergence of BCH and BSV is caused by the disagreement of big miners.
3) Another problem is that this kind of mortgage will greatly reduce the circulation of FIL. Therefore, some people say that miners and miners will be the most important backing for FIL. What do you think?
Li Bai:Miners are inseparable from miners. Many miners themselves are also big miners and involve many cooperative relationships, so the interests of miners and miners are the same. Filecoin mining involves a lot of optimization, operation and maintenance, and hosting, all of which are inseparable from miners, which is different from POW mining.
As for the miners' right to speak, I don't think it is a problem. The miners are not monolithic, and they are also playing games internally. If some people fork to another chain, then the computing power competition of the original chain will inevitably ease, which may be a good thing for the remaining miners.
As for community splits, this is more complicated. Technically, the fork is very simple, because the code is open source, as long as you have mining machine resources, there is no problem, but the developer community, user base, follow-up development, etc. cannot be easily copied, which is why there are many bitcoins There are many forked coins, but there are very few forks in Ethereum. Miners will certainly consider these issues carefully before acting.
As for whether the miners will be the currency price support, I think the miners themselves will choose. In essence, it is still a matter of calculating whether the cost is cost-effective. It also depends on the economic model and cost at the time.
Reader's question Q1: Some nodes have high computing power, but the success rate is very low. What is the reason?
Li Bai:There are many factors in the order success rate, because the official order-issuing robot is in Europe, the miners are in China, and the direct connection network environment between China and Europe is not very good, so it is necessary to go through an agent. If the settings are not good, this is a big hurdle; in addition The most important thing is network optimization, order priority optimization, these internal system issues.
Reader question Q2: Is there no reward if the success rate is lower than 80%?
Li Bai:Rewards are on the one hand, the official rewards for this competition are quite generous. But on the other hand, this is also part of the test, which is a simulation of the situation after the mainnet goes live. The design of filecoin is relatively innovative, and the project team and miners are trying to find out from each other. Only after experiencing these tests will they have more confidence in the main network.
For the reward this time, the two 80% are the threshold for getting the reward, if you don’t reach it, you don’t have it. But maybe, the official may change it again, after all, they are the ones who give out the prizes.
Readers ask Q3: What do you think is the optimal configuration of a cluster size, how many computing devices and how many storage devices should be allocated.
Li Bai:Different configurations are different. We also sell clusters now. The cluster size we recommend for 7-series machines is 48 computing machines. The storage depends on your needs. The cluster solution we provide now is a single package with 3T effective computing power, so the daily effective computing power of a cluster is 48*3=144T.
Odaily：Okay, thank you Li Bai for your wonderful sharing today, and I look forward to your follow-up wonderful performances.
