Recently, the distributed storage track that has been popularized by the Filecoin mainnet has come into people's field of vision.
While investors are studying Filecoin's mining scheme and token model, they are also concerned about the feasibility and growth of decentralized distributed storage.
In order to help new investors get started quickly and improve their awareness, on July 22, Odaily held the "Interstellar Rise - IPFS Technology and Distributed Storage Ecological Cloud Summit", and discussed "Distributed Storage" in the fourth round table business future:
Economic Model and Interpretation of Cycle” has made an in-depth discussion.
The roundtable was hosted by Zhang Chao, partner of BiNiuNiu, and the guests included Danfeng Lou, founder and chairman of Diancun Technology, Calvin Cao, co-founder of Starland, Yang Xiao, COO of Panda Miner, Lao Feng, co-founder of Mirror Network, and Yan Zitong, founder and chairman of Xi'an Lingdong Computer System Co., Ltd.
In the round table, the guests discussed the market space of distributed storage, the trend of decentralization and monopoly of distributed storage, and the industry opportunities of distributed storage outside of the mining industry.
Lao FengA case of failure of Huawei's decentralized distributed storage project was cited to illustrate that the emergence of blockchain super ledgers has brought the possibility of truly decentralized distributed storage, which allows various enterprises and institutions Take out idle resources to participate in a super-large-scale distributed network with confidence. The most important thing is that the process is clear and all data is uploaded to the chain. Otherwise, distributed storage will be difficult to achieve.
Yan ZitongComparing Filecoin with the distributed storage technologies of Amazon, BAT and other service providers, it believes that BAT and other manufacturers are self-operated, while Filecoin is similar to the "Didi Taxi" of distributed storage. By providing a platform, contributors can provide Freely price your own services to provide users with more choices. And because of the addition of the token economy, contributors who are optimistic about Filecoin for a long time can set a low or even loss-making price for the service, and subsequently obtain positive benefits from the growth of the network and the value-added of FIL.
Lou DanfengResponded to issues such as the high threshold for early participation in Filecoin and the risk of monopoly. It said that in the early days, for the sake of network security, Filecoin raised the threshold for participants from mining mechanism to economic design; in addition, because Filecoin provides anonymous content storage and information retrieval, it is likely to bring harsh content Therefore, the network is unlikely to be used on a large scale like the centralized storage network in a short period of time. At this stage, there will inevitably be a certain concentration of computing power and tokens. The official is reducing monopoly by unlocking in batches, staking by miners, and accelerating the implementation of applications.
Calvin CaoForecast the supply and demand of the storage and retrieval market after Filecoin goes online. He said that the price of the Filecoin storage market is determined by the relationship between supply and demand. The storage demand is small and the number of miners is large, so the storage price will naturally fall. In the early days, the amount of effective data was smaller than the number of miners, and the supply in the storage market was greater than the demand, resulting in extremely low storage prices, which attracted many companies that needed low-cost cold storage to test the waters, but cold storage did not have a large demand for the retrieval market, so This market will not have a large volume at the beginning.
Yang Xiaosecondary title
The following is the essence of the roundtable discussion, organized by Odaily:
Chao Zhang:First of all, I would like to invite all guests to briefly introduce themselves and the company's main business.
Old Feng:Hello everyone, I am Lao Feng from Mirror Network MW. I entered the circle in 2012. I have been in the bitcoin field for three or four years, engaged in the production and development of chip equipment. In 2016, we began to develop a decentralized distributed storage system based on the IPFS protocol. This system was formed last year, which is mirror image Network MW.
The goal of the mirror network MW is to become an available and easy-to-use blockchain + distributed storage network. At present, there are nearly 8,000 storage nodes in the entire network, and the available storage space of the entire network is close to 260PB, and the storage of data in the network has been realized. In the future, we will continue to establish nodes, expand this network, and realize security, privacy, and very cheap and convenient storage fields.
Lou Danfeng:Hello everyone, I am Lou Danfeng, the founder of Diancun Technology. Diancun Technology entered this track in 2018, mainly to provide solutions for mining machine manufacturers. In addition, there is a company called Byte Ark, which is doing edge computing business. Cooperation, such as State Network ICT, ByteDance, iQiyi, Kingsoft Cloud, Yunfan and so on.
Yan Zitong:Yang Xiao:
Yang Xiao:Hello everyone, I am Yang Xiao, the co-founder of Panda Miner & Niubit. Founded in 2014, Panda Miner is the first batch of digital currency miners in the world. The current business layout covers local mining hardware, mining infrastructure, miner services and ecological expansion, covering the global miner community.
Niubituo, a subsidiary of Panda, is one of the oldest cloud computing power platforms in the world. Relying on strong supply chain resources and rich operating experience, it has provided safe and reliable cloud computing power mining for users in more than 50 countries around the world. Serve. This year, we launched the Panda IPFS one-stop solution to lay out the next 100 billion market.
In this regard, I think Panda's core strengths lie in hardware supply and IDC computer rooms.
As the No. 1 brand of graphics card mining machines, we have accumulated a large number of partners in the mining machine hardware supply chain, such as AMD, Samsung, and Intel. We can get goods directly from the brand, or even start production by ourselves. When building mines in the early stage, we also accumulated a large number of low-cost and high-quality IDC computer room resources in Inner Mongolia, Guizhou and first-tier cities. Coupled with years of operation and maintenance experience and service awareness, it can meet everyone's IDC computer room hosting needs. If you have appeals for component procurement, machine mass production, and IDC hosting, just come to us directly.
Calvin Cao：Hello everyone, I switched from the operation and maintenance of overseas mines in the traditional Bitcoin mining circle to the current IPFS circle in 2014-2015.
Our Interstellar Continent was established in April 2018, and the team members are former executives from first-tier companies including Huawei. In the past two years, we have mainly been doing technical optimization, including the improvement of some algorithms. We are still running well on the P1 and P2 algorithms. On the SDR, we can also achieve more than twice the official speed. The foothold of the mining machine Finally, the input-output ratio is above.
In addition, our blockchain browser has been made, and our mining software and mining pool contracting system will be announced in the future. You can pay attention to it.
Chao Zhang:Let me also briefly introduce. I am Jane Business VC, a partner of BiNuNiu. BiNiuNiu is positioned as a "smooth flush" in the field of encrypted digital assets. It is committed to providing users with the latest market and information. At present, our Android download volume exceeds 8 million, and the highest daily activity exceeds 150,000.
Our team is relatively strong. The core personnel come from well-known companies such as Microsoft, Tencent, Huawei, Oriental Fortune, and Flush. They have more than ten years of background in Internet and finance.
We have also received a lot of favor at the capital level. Since May 2018, BiNiuNiu has obtained strategic investment from De Capital, Ouke Group, Huihui Fund, Bixin Capital, and Genesis Capital. Tens of millions of investment such as OKlink.
I believe that everyone should be very sure about the future of distributed storage, including the trend of computing, storage and other network resources from centralization to distribution, as well as the combination with future new production relationships (sharing economy, contribution-reward fragmentation) . For the first question, I would like to ask the bosses to roughly estimate the growth rate of distributed storage in the overall storage market, and briefly talk about the basis behind it.
Lou Danfeng:Yes, the growth rate of distributed storage in the overall storage market will be very fast. Let's look at the following points:
First, rigid demand.
According to IDC statistics, the total amount of our global data will reach 44ZB in 2020, and the global data will grow to 175ZB by 2025, which is more than 5200GB per person. The scale of the global storage market has reached more than 6 trillion US dollars, and 80% of the market is occupied by China. The performance and capacity of traditional storage are inversely proportional, and the ability to store massive unstructured data is limited. Distributed storage can achieve high scalability and easy management, so I believe that in the future, more and more enterprise data will choose distributed storage. With the rapid development of 5G, the Internet of Things and artificial intelligence, by 2023, About 40% of the world's storage systems will be distributed storage architecture.
The second is for enterprise-level needs.
Enterprise data will become the mainstream of my country's data. In 2015, enterprise data accounted for only 49%, and in 25 years it will increase to 69%. Traditional information systems are dominated by CRM, ERP, and OA, and the data scale is small; cloud computing, big data, Internet of Things, and AI new technologies are widely applied to enlarge the data volume. At this time, most enterprises still use traditional storage methods and cannot To effectively store and manage PB-level data in future enterprises, enterprise-level storage urgently needs a transformation of distributed storage.
Yang Xiao:
Yang Xiao:Mr. Lou above also mentioned the problem that the centralized server has a load limit. In fact, the centralized computing method has a series of disadvantages:
The limitation of centralized single-point resources and complete loss of control after failure;
Resource monopoly and cost increase due to supplier lock-in;
Traditional centralized databases are intensive and unique, which brings potential risks to data privacy.
For these problems, distributed computing and storage reduce the heat loss of data centers on the one hand, and at the same time make it easier for data providers and consumers to obtain the required computing resources. On the other hand, blockchain-based distributed computing can use the contribution proof protocol to activate distributed computing resources through token incentives.
According to IDC's forecast, with the rapid development of 5G, the Internet of Things and artificial intelligence, by 2023, about 40% of the world's storage systems will be distributed storage architecture.
Another research firm, Zion Market Research, predicts that global distributed storage sales will grow rapidly from US$2 billion in 2018 to US$28.5 billion in 2027, with an annualized growth rate of 34.7%.
The global market is optimistic about the future of the distributed storage market, which is the general trend.
Calvin Cao：At present, our decentralized storage is a combination of distributed software + server, which has several advantages:
High scalability, large number of nodes, can achieve EB level expansion;
Operation and maintenance is more convenient than centralized storage, and file types such as blocks, objects, and files can be managed in a unified manner;
Hardware standards are relatively uniform.
For example, if we compare cars to storage and data to goods, the previous centralized storage is equivalent to a van, and the capacity for pulling goods is limited. The current distributed storage is similar to a locomotive. If you want to pull more goods, you can directly add compartments , The carriage also has power, so you are not afraid that the train will not move.
The previous guests also mentioned that regardless of the sales data in the past few years or the sales forecast in the next few years, the spring of distributed storage has arrived. Open source distributed storage represented by Ceph is booming, represented by EMC and Huawei. With the heavy investment of professional storage manufacturers, I believe that the combination of IPFS+Filecoin with its own incentive layer will develop better and better in the future.
Yan Zitong:We know that mature cloud service providers like Amazon and BAT are already using distributed storage technology. In this context, what competitiveness does Filecoin have in doing this?
Let's first look at the essential differences between the two. Filecoin is similar to Taobao, Meituan and Didi. It provides a platform to connect contributors and users, which is different from the self-operated model.
What is its impact?
Suppose I want to store a file now. For example, in Alibaba Cloud, the quotation is 0.3 yuan/month, which is clearly stated. On Filecoin, the miners who provide this service are competitors. In addition, the entire network uses tokens as the value carrier, which is different from the currency standard. The legal currency is one dollar today, and it will still be one dollar tomorrow, and it may depreciate slightly a year later, while FIL has a potential upside.
Then some people who are optimistic about the development of Filecoin may quote a price at a loss based on the legal currency standard. He gets the order, the network becomes more practical and robust, and the value of FIL rises, and he can also profit from it.
Therefore, Filecoin is a change in the way of productivity, which will change the current situation of high energy consumption of distributed storage operated by centralized manufacturers.
Old Feng:I would like to share a case first. In 2013, Huawei set up a data center overseas, which brought together small storage operators from different countries. why? Because these operators are located in countries with poor infrastructure, fragmented networks, and high idle rates, it seemed like a good idea to pool their storage capacity and sell it, but it failed. Why, the reason is that there is a problem with its incentive mechanism, and the amount of work has not been rewarded accordingly.
We believe that the emergence of blockchain super ledgers has brought opportunities for truly decentralized distributed storage. It allows various enterprises and institutions to take out idle resources to participate in a super-large-scale distributed network. , the most important thing is that this process is clear and all data is uploaded to the chain.
Distributed storage is the future trend, and the biggest point is to provide users with near-real-time data services. The current 5G and future ultra-high-speed networks certainly cannot all rely on centralized servers. Centralized servers can provide intelligent distribution of applications and processing of AI artificial intelligence. The calculation and storage of data will definitely be handed over to distributed data centers. , The two exist at the same time. I think that in the next five to ten years, society can transition to such an era of multi-centered distributed data services, and it will become the mainstream.
Yang Xiao:
Yang Xiao:Regarding the threshold, first of all, Filecoin hopes to build a sharing economic model to solve the problem of resource misallocation. But in the early stage, for the stability of the network, and because the technology could not do it, the next best thing was changed to using professional storage mining machines to store things. This is time-sharing leasing.
However, I think that the overall gradual decentralization will be the development trend of the entire distributed storage, which may eventually go through three stages:
sharing economy. Anyone with a computer and a mobile phone can become a miner as long as they are willing to contribute storage space, realizing truly distributed storage.
Timeshare lease. Mining machine miners can receive orders with real storage needs in the market through the upgraded order receiving system;
sharing economy. Anyone with a computer and a mobile phone can become a miner as long as they are willing to contribute storage space, realizing truly distributed storage.
Of course, it cannot be said that the current first and second stages are meaningless. After all, the sharing economy like Didi and Mobike also started gradually from time-sharing leasing.
Old Feng:I think that there is no problem that individuals cannot participate in distributed storage projects. One is the difference in the products he invests in. Now there has been a cut in computing power and the emergence of financial derivatives. Our best configuration is also Just over 20,000 yuan. In the end, distributed storage must be a giant network, and even multi-chain and cross-chain architectures will appear to combine the resources of all distributed storage nodes.
Lou Danfeng: I don’t think so. In addition to the initial security issues mentioned by the previous guests, because the Filecoin network has the characteristics of anonymity and privacy protection, the subsequent content storage and information retrieval are likely to bring To more stringent content control. Therefore, in a short period of time, this network is unlikely to be used on a large scale like the centralized storage network.
In addition to unlocking in batches, is there any effective way to slow down the speed of monopoly?
First of all, here is why it is said to be unlocked in batches. The Fillecoin project is to be a blockchain storage landing project. Mining is actually to obtain the coinage right. If the coinage right is in the hands of a small number of people in the early stage, and released to In the market, the currency price is at risk of being manipulated, so the probability of unlocking in batches is very high, and it may even be very high in the early stage, no matter from the perspective of miners or the source of crowdfunding rounds in the early stage. Pretty high.
In addition, other methods are to control the currency price or the way miners pledge, to accelerate the implementation of applications, and so on.
Calvin Cao：What I would like to add is that in addition to unlocking in batches, Filecoin also adds a baseline network benchmark, which is not included in the Filecoin white paper.
The purpose of adding this amendment to the Filecoin system is to prevent the unfair situation of a small number of miners obtaining too much Filecoin in the early stage, and to delay the issuance of new coins when the computing power of the entire network is relatively small.
The Baseline reward model has two benefits for the Filecoin network:
Linking block rewards with the growth of the storage market avoids the unfair situation that miners quit the network after receiving large early rewards, and also prevents some miners from monopolizing the market due to excessive rewards in the early stage.
Smooth the exponential decay model to avoid large-scale changes in the storage capacity of the entire network due to reward halving.
Yan Zitong:Considering the security of the initial stage of the network, billing and other issues, the project party finally canceled the VDF certificate and switched to zero-knowledge proof. Everyone will feel that such a high configuration and AMD components that are difficult to purchase, so the threshold is too high high. But if the early security is poor, it is very likely that it will fall as soon as it comes out.
I believe that with the increasing perfection of the Filecoin network, the increase in the number of nodes, the improvement of the security of the storage network, the continuous improvement of the IOO capability of the network, and the increasing demand for real data storage, more commercial companies Similar to us and Lao Feng’s MW, when added to the development of ecological applications, the usability of Filecoin will be higher and it will become within reach.
Zhang Chao: The last question, when we talk about the distributed storage ecology, it is more from the perspective of the mining industry. After the main network is launched, will the storage demand side be further subdivided in addition to the mining industry (supply)? What service-oriented roles may be involved in distributed storage, and are there any industrial opportunities other than mining?
Calvin Cao：The price of the Filecoin storage market is determined according to the relationship between supply and demand. If the demand for storage is small and the number of miners is large, the price of storage will naturally fall; if the demand for storage is large and the number of miners is small, the price will rise. What we are talking about here is not currency price but Don't get confused with prices in the storage market.
The current market is mainly an input and output market, that is, a storage and retrieval market. In the early stage, when the amount of effective data is less than the number of miners, the supply of the input market is greater than the demand, resulting in extremely low storage prices, which will attract many companies that need low-cost cold storage requirements to test the water, but cold storage does not meet the demand for the output market. Large, so the search market will not have a large volume at the beginning.
At present, IPFS has been widely used, and now traditional Internet giants including Google, Firefox, Amazon, JD.com, and Alibaba Cloud are using IPFS. And mainstream blockchain projects such as Ethereum, Polkadot, and Ontology have already used or are bridging IPFS and Filecoin. I think the opportunities outside the mining industry are mainly in the secondary market and the development of Dapp.
Yan Zitong:I think, no matter from the perspective of enterprises or individuals, large enterprises have huge demand for data storage and backup, but they will face a direct problem, that is, they cannot pay directly with FIL. Taxation is a big problem. Then at this time, a third-party intermediary service agency and consulting service agency may be born, which can convert the legal currency paid by users into FIL.
Old Feng:In terms of storage requirements, the mirror network MW positions itself in the cold data industry. We believe that building a cold backup network is the most suitable scenario at this stage. Our biggest advantage is hard disk capacity. Computing and network are not advantages. As the security, speed, and scale of the network change, it will definitely touch various industries.
The biggest advantage of distributed storage should be that it is cheap. While acquiring coins, miners can also rent storage space to those in need, so how to sell these spaces is more important. I believe that there will be a lot of application-level products. There are many service roles here, such as operators, custodians, technical operation and maintenance, storage products, DAPP commercial companies, etc.
Lou Danfeng:At present, the entire mining machine market and the industry itself are close to the risk point. According to incomplete statistics, there are at least 500+ large and small mining machine manufacturers and sellers. The sales of the large ones have reached billions, and the small ones are hundreds. From the fact that the sales volume of the entire mining machine has reached the level of 100 billion, the total value of server sales has far exceeded the total value of Filecoin at the time of private placement, because Filecoin was mainly obtained by block rewards in the early stage, and the real application data There will be very few, and there will not be that much liquidity in the short term. As a result, some institutions and markets now estimate that the currency price will range from 15-30 US dollars per coin three months after the Filecoin mainnet goes live. Looking at it now, the currency price plus the uncertainty of the economic model, the Filecoin project has a lot of risks. Specifically, there are several aspects:
The Filecoin project has bounced many times. Many manufacturers and investors have reached a critical point in the credibility of the project. The investment in research and development costs, testing costs, and a serious waste of a lot of resources have caused many manufacturers to dare not start production at once. Judging from the current test network There are only tens of thousands of servers in operation, worth more than one billion yuan;
The supply chain in the server market is tense and prices are rising, especially in the past two months, some accessories have increased by more than 100%;
Filecoin’s official changes are frequent, causing some manufacturers’ technology to fail to keep up, and some even give up this track directly.
Based on the above, in fact, many manufacturers have begun to deploy business outside the Filecoin industry to hedge against some uncertain risks of Filecoin, such as Ai computing power platform, GPU outsourcing service, PCDN, rendering, cloud service and other businesses.
Of course, the track of Filecoin is also extremely competitive. After the mainnet is launched, the tide recedes, and there is no shortage of companies with technical strength and delivery capabilities. Those with algorithm optimization capabilities even exceed the performance of some companies by several times. The final competition is several factors such as equipment stability, safety, and efficiency.
In addition, regarding the upstream and downstream of the distributed storage industry as a whole, there are actually great opportunities. You can do a whole set of solutions, you can also do algorithm optimization, or some application levels, or you can do something from this track. The solutions that are good at in the field or the outsourcing of computer room operation and maintenance, etc., these are some demand points around this industry.
For example, we click to save, and now we can provide some large customers with the entire mining capacity output solution, from inability to capability, from the customer's original output of 1 production capacity to 2 production capacity, including hardware level selection, algorithm optimization, network solution customization, operation Dimensions, security protection. Now we cooperate with some big companies, such as Alibaba Cloud, Mellanox, Huawei, Lenovo, Cisco, and Dell. Only with such a complete technology combination can you make better solutions.
At the same time, more than 50% of our company's technical personnel are from some large factories of traditional cloud service companies, and about 20% of our technical personnel are from the blockchain industry.
Yang Xiao:
Yang Xiao:Just now, the bosses have explained the entire ecology relatively well. Finally, I would like to add on which dimensions and fields the entire application ecology may explode. I think at least these four aspects can be seen.
First, the security and privacy of information; such as family camera data, such data involves privacy data, the optimal way is not a centralized storage server.
Second, the permanent storage of information; such as medical information files, astronomical data and scientific research data, such cold data may be lost in three to five years if there is a centralized server, and it will be permanently stored if it exists in the Filecoin network.
Third, the scenario where data is naturally distributed, blockchain, edge computing, and the Internet of Things, are naturally compatible with Filecoin's distributed storage.
In terms of blockchain, the essence of blockchain is a distributed ledger, and one of its bottlenecks is the storage capacity of the ledger. At present, the biggest problem of most public chains is that they cannot store a large amount of hypermedia data on their own chains. The total block data of Bitcoin so far is only about 30-40G. Programmable blockchain projects such as Ethereum can only execute and store small pieces of contract code. DApps want to develop into super apps, which is greatly restricted.
Fourth, confirm the rights of the content; for example, data transactions must first confirm the rights of the data. The reason why data is not dared to trade at this stage is that after one transaction, it can be copied infinitely and lose value. After the data rights are confirmed in the future, it can be done Data value.
Chao Zhang:Okay, thank you for your sharing.
