current position:Home>Data on human: decentralization of cloud storage

Data on human: decentralization of cloud storage

2022-08-03 12:10:05Sina Finance-Blockchain

  文章作者:Hunter Lampson 金色财经

  Article reproduced with permission of the author,作者Hunter LampsonGoldman sachs analyst,Interested in capital deployment and digital assets,Will be in the creator community over the weekend,Can be found on twitter he.

  Data determines the human.For technology innovation and the pursuit of a human life digital,Created a demand for data storage and retrieval of explosive.From the agricultural revolution、Healthcare Discovery and Political Archives,To self-driving cars、Protein Folding and Neural Networks,Data is to help us find the main booster achieve the goal of the new solution.It is the basic tool,Restrictions and forced us to action ability and the agent shall not reduce investment,Allow access to and give our digital and physical meaning of life.Data on human:We must be very concerned about how our data storage、管理和拥有.

  Global data market

  今天,超过63%的全球人口,即50多亿人(Google data shows there77亿人口),使用互联网,This number will continue to increase every year10%以上的速度增长.But the cloud storage market is growing faster,从2015年到2025年,Global data field(On a global scale to create、捕获、Copy and consumption amount of data)预计将以58%的复合年增长率增长,到2025年,创建、The amount of data storage and replication will be over180ZB(1 zb等于1024tb,1tb等于1024GB).如果你在2025Years before stack enough10tbHard to meet the demand of data around the world,Then this pile of hard drives can reach the moon.

图 1.  Global datasphere size by year.资料来源:Uygun 和 Döngül,2021 年图 1.  Global datasphere size by year.资料来源:Uygun 和 Döngül,2021 年

  从经济角度来看,Cloud storage market in2021In the valuation of about760亿美元;到2028年,它将达到3900亿美元(年复合增长率为26.2%).despite such explosive economic growth,Cloud storage providers market share continues to consolidate.截至第22季度,三大云提供商——亚马逊网络服务(AWS)、微软Azure和谷歌云平台(I affectionately call it cloud services three giants)——Now the cloud computing market65%的份额.Centralized power of cloud storage providers have integrated their network effect、声誉、Technology Infrastructure and Balance Sheet,That new competitors cannot compete with.

  1. 本地

  2. 集中式(中心化)云存储 (CCS)

  3. 分散式(去中心化)云存储 (DCS)

  本地存储和CCS提供商——三大巨头(Amazon、Azure、谷歌)And alibaba cloud、Box、iCloud等——They are based on the characteristics of centralized storage way.This means that information is stored and maintained in a single location(Or the minority position),Management in a single database,Operated by a single entity and,Local deployment andCCSThere is a single point of failure risk solutions.

  CCSThe popularity of solution need to be done to the on-site data storage efficiency of history.起初,Users store data on their own hardware.This means that the data storage and maintenance of all hope entities stored in the same physical location(For example the company's existing data server),I call it the first stage.

图 2.  Three Stages of Data Storage Adoption.资料来源:Hunter Lampson图 2.  Three Stages of Data Storage Adoption.资料来源:Hunter Lampson

  As the cloud storage network effects for cheaper(Usually more safe)的存储能力,Consumers and companies move to centralized cloud(第二阶段),CCSSolutions developed for cloud computing、API和其他SaaS产品,The customer also grow.Although the centralized solution is on the market the most simple、The choice of the cheapest and most effective,But their basic limitations remain the same:A container is responsible for the entity's100%数据.CCSThe solution is an improvement over the indoor solution,But once the economy best solution has become expensive and a forbidding place.今天,DCSProvider is the cheapest on the market and the most secure storage solutions.

  1. The lack of data ownership

  当用户向CCSProvider when uploading data,They no longer own their data.Apple scanningiCloudThe user's controversial decision(Later revoked)就是一个很好的例子.When data is stored on a given hardware product(iPhone、Mac等)上时,Apple has a strict privacy policy.但重要的是,一旦用户向iClouduploaded one byte of data,Apple will think these data belongs in the realm of their——and no longer belong to the user's domain.This precedent means,Data stored locally belongs to the user,The data stored in the cloud are storage providers.

  2. prone to data breaches and outages

  在CCS供应商中,A large amount of data reveal that does not need to spend too much time.Amazon、Azureand Google due to its single point of failure structure,All suffered from this problem.

  The centralized structure of these providers allow them to build large walled garden,And provide relative to the internal solution for higher level of security.与此同时,The greater the database becomes、越集中,The attacker more coveted it.Data interruption inCCSIn the solution is also very common,Examples can be seen here:Amazon, Azure,谷歌.

  3. Tend to censorship

  CCSProviders not only lose data uncontrollably,But also deliberately to delete data.就在几周前,YouTube热门频道BanklessWithout any warning、terminated without notice or reason.Google owns and stores on its cloud serviceYouTube内容,谢天谢地,It restored the channel,But Google and otherCCSThe provider had to remove the existence of certain data,It is harmful to society.

  4. 高成本

  也许CCSThe most critical disadvantage of the solution is that it is too expensive.尽管在过去的50年里,The cost of storing data average annual decline30.5%,但CCS的价格在过去的7Years remains the same.这是由于CCSNetwork effects accumulated by providers.As a result of these network effect,The big three have begun to dominate the field of cloud computing.As their joint market share continues to grow,The big three as an oligopolistic enterprises,Have the ability to manipulate the price,and shut out new entrants.

图 3.  Data storage cost over time,资料来源:Arweave 黄皮书图 3.  Data storage cost over time,资料来源:Arweave 黄皮书

  图 4.  AWS、Azure、Google Over a period of time on data storage costs.资料来源:AWS、Azure、谷歌、Hunter Lampson.

  Price and storage cost is the main reason for the gap betweenCCSProvider currently in the market dominant position,DCSThe solution took a different path.


  在CCSThe weakness of the above,去中心化存储(DCS)Has been shown is a paradigm shift in the field of data storage(第三阶段).DCSSolution by matching the storage space of supply and demand,Using node set across the geographic distribution of free disk space.This creates a more efficient market,降低了成本,To eliminate local andCCSRisk of a single point of failure in the solution,DCSSolutions will be returned to the user data ownership.

  图 5. According to the platform storage of the year 1 GB The cumulative cost.资料来源:AWS、Azure、谷歌、Storj、SiaStats、Arweave Fees、、Hunter Lampson.

  Although data center and storage nodes distribution is not the only factor in determining that person's network concentration,But it's a useful touchstone.Across the space of node distribution is also decided to copy、Retrieve and protect the data level of an important factor.一般来说,The more nodes in a network of,The faster the speed of retrieval,The protection of natural disaster is the stronger(When are we going to put storage nodes on the moon?!)因此,Understand decentralization is the prerequisite for effective cloud storage nodes is very important.

  与CCS解决方案相比,DCSSolution of revolutionary place lies in its decentralized degree.在Sia、Storj、Filecoin和ArweaveActive nodes running onAWS、Azure和谷歌CloudManagement of the data center combined114倍以上.

  图 6. Total Active Nodes by Service.资料来源:Filscan、Viewblock、Storj、SiaStats、Peterson 2015、Baxtel、谷歌、Sam Williams、Hunter Lampson.

  ArweaveThe number of nodes are hard to quantify,因为ViewblockThe statistical data provided by each storage pool as a single storage node.In an offline dialogue,Arweave的创始人Sam Williams告诉我,59The current storage pool(根据Viewblock)Hundreds or even thousands of nodes can support them.因此,ViewblockUnderestimate the actual number of nodes10-100倍.出于这个原因,为了保守起见,我使用“500+”作为节点数.还需要注意的是,Active node count is decentralized imperfect measure,The absolute number of nodes does not tell us who is running the node(and how many nodes each entity operates).

  借用Spencer Applebaum和Tushar Jain的说法,DCSService is one of the most important difference between storage solutions based on the contract and the difference between permanent storage solution.简单地说:On the market at present all of theDCSServices are all contract-based models,只有Arweave例外.

  The storage model based on contractsvsPermanent storage model

  Filecoin、Sia和StorjUsing a contract-based pricing model——与目前CCSThe company uses the same model.Store data based on the contract price means the user continue to pay the cost of,Similar to a paid subscription(每个月/Every year to pay).尽管存在细微差别,但Filecoin、Sia和Storj与现有的CCSProvider of direct competition.

  另一方面,ArweaveProvides a permanent storage model,This means that users only need to pay a fee,作为回报,Their data will be stored permanently.与其他DCS和CCSProviders than,ArweaveOften appear lazy and imprecise,ArweaveThe basic characteristics of different from competitors is data persistence.

图 7. CCS 和 DCS Solution concept map,资料来源:Hunter Lampson图 7. CCS 和 DCS Solution concept map,资料来源:Hunter Lampson

  仔细研究Filecoin、Sia和Storj,help us better understand them andCCS提供商和Arweave的不同之处.

  图 8. DCS The main characteristic of the solution.资料来源:Filecoin、Storj、Sia、Arweave、CoinMarketCap、Crunchbase.

  Filecoin于2020年10月推出主网,Is the most widely used on the market at present、The most money enoughDCS项目.截至2022年7月12日,FilecoinMarket value of the fully diluted about11.9亿美元,Record for123亿美元.Juan BenetLaboratory is agreement(Protocol Labs)的创始人兼首席执行官,该公司开发了FilecoinAnd underlying technology——星际文件系统(IPFS).到目前为止,Filecoin已经筹集了2.582亿美元的资金,其中大部分来自2017At the end of the year for the first time issued tokens(ICO).

  要理解Filecoin,我们必须理解IPFS,This is a used to store and retrieve data point to point(P2P)分布式系统.IPFSIs to solve the problem based onhttp的internetFaults and build,It USES the content addressing to classifying data,This means that the information is based on the content rather than its position to request and pass.This is by an issue of a content for each data block identifier(CID)来实现的,CIDIs through the content of each file hash generated,使其不可变.In order to locate the requested information(由唯一CID表示),IPFS使用分布式哈希表(dht), dhtContains the storage andCIDThe network location of the node of the relevant content.当用户从IPFSRequest information node,The node will check its own hash table,See if you can locate(然后检索)所请求的文件.If the node does not contain the requested information,It can be downloaded from the peer node content and deliver it to the user.在这个模型中,Information is copied across multiple nodes,而不是HTTPModel of single、集中的位置.This eliminates the risk of a single point of failure,At the same time improve the retrieval speed,Because the data is from multiple peers retrieval at the same time.

  IPFSis a communication network used to store and transmit data,而Filecoinis the economic system built on it.IPFSIn and of itself can't motivate users to store the data of others:Filecoin可以.This is proved by two unique mechanism of:复制证明(PoRep)和时空证明(PoSt).PoRep只运行一次,To verify whether storage mill has what they said.For each of the chainPoRep,都包含10个SNARKs(简洁的非交互式知识论证),This proves that the completion of the contract.另一方面,PoSt是连续运行的,To show that storage miners will be stored in a period of time space dedicated to the same data.Verify that the process of interaction required chain's data intensive,因此Filecoin使用zk-SNARKs(零知识简洁非交互式知识论证)To generate these prove,And its data compression to10倍.

  In the discussion of four kinds ofDCS协议中,SiaIs the first release of,并于2015年6月发布.2013年,David Vorick和Luke Champine在HackMIT创立了Sia,The company has strong user appeal,The fully diluted market capitalization reaches1.9亿美元,History of the highest29.7亿美元.

  Sia是由成立于2014年的nebula Labs推出的.Sia以类似于FilecoinTo upload data into composite parts(In this case, the fragment),And spread them to the distributed hosts around the world.与Filecoin不同的是,SiaThrough the different storage prove(PoS)Mechanism to realize it.This proof for the host with the passage of time sharing a small group of randomly selected data.The proof is verified and stored inSia区块链上,主机获得Siacoin奖励.

  与Filecoin和Sia一样,Storj自2018年10has gained huge traction since its launch.Storj与Filecoin和SiaDifference is that it does not rely on block chain consensus to store data.相反,StorjTotally dependent on node to store data erasure coding and satellite,to increase data redundancy and reduce bandwidth usage.StorjExclusive use of erasure coding means data persistence(The probability that the data will still be available in the event of a failure)With the expansion factor(The additional cost of storing data reliably)不是线性相关的.因此,在Storj上,Higher persistence does not need to scale up bandwidth.Switch a given node(Nodes offline(Or leave the network)的速率),从长远来看,Erasure coding can be valuable,Because it requires less disk space and storage and the bandwidth of the repair,Although it increases theCPU运行时.

  StorjIn terms of network architecture and pricing mechanism, it is also related toFilecoin和Sia有所不同.在Storj上,Pricing by the intermediate storage user(包括应用程序)And the satellite node of the storage node decides.Satellite node is responsible for the negotiated price and the bandwidth utilization.因此,StorjPricing model is not completely depend on the free market activity,But subject to concentrate on,Because satellite operators represents a potential concentration of intermediary between a node and the end user.

  Storj还与Amazon S3进行了本地集成,这意味着现有的Amazon S3The user can migrate toStorjBasic features and use,without changing their codebase,This may be reduced and leaveAmazon S3Ecological system related friction.

  与Filecoin、Sia和Storj不同,ArweaveProvide permanent data storage.Arweave由首席执行官Sam Williams和William Jones于2018年6月成立,截至2022年7月12日,ArweaveThe fully diluted market value to8.9亿美元,达到41.8亿美元的历史最高市值.

  ArweaveSeek to decentralized way provide permanent data storage,Once to pay,这是通过 Arweave Donation mechanism complete.Considering the cost of data storage in the past50Years per year30.5%的速度下降,ArweaveThink every todayGB/1The purchasing power of dollars in storage is greater than the futureGB/1Storage costs are higher for $.The delta madeArweaveThe donor pool of possible,“本金”is the upfront fee paid by the user,“利息”Is the purchasing power over time tokens, prices.ArweaveA conservative assumption is that storage prices fall annually0.5%,This enables the donor pool to long-term survival.

  Arweave 目前约 3.85 美元/GB The cost reflects the end value of data storage.在短期内,Sia 和 Filecoin(甚至是Big3三巨头)更便宜.但从长远来看,Arweave Become more wise choice.即使在短期内,Users will pay a premium for what others cannot provide:数据持久性.对于某些人来说,The cost of permanent storage is relatively inelastic,Because some files(例如 NFT)需要永久存储.

  ArweaveBy the support block woven,It is a data structure similar to blockchain,Each piece had links to the previous block area and recall block.Recall block is except the piece of recent mining is mining before any piece of.因此,ArweaveStructure is not only the continuous chain block linked together——It is the current link to the previously dug out the block and another random block(Recall块)Weaving together.

  To mine new blocks and get mining rewards,Miners must prove that they can access recall block,ArweaveAccess to prove(PoA)机制保证,For each newly mined block,Data for recalled blocks is also included.This means storing new data,Miners must also store existing data.PoAAlso encourage miners equality between the nodes to copy all data.When selecting with low levels of replication as a recall of the block,The men will be able to use this block in less miners pool competition the same reward.在其他条件相同的情况下,随着时间的推移,With low levels of storage replication blocks miners will gain greater reward.

  Built on top of the block weave is the Permanet——Similar to the world wide web today,But a permanent.Arweave的Bockweave是为PermawebPower base layer;PermawebIs the user interacts with the layer of.鉴于Arweave是建立在HTTP之上的,Traditional browsers can access to the network of all data stored on,resulting in seamless interoperability.


  虽然DCSThe solution could theoretically be better thanCCS解决方案,But should be based on their usefulness in practice to evaluate them,We can check the following content to measure the attractiveness of each project:

  1. 存储的数据

  2. 节点分布

  3. Interested in search

  4. 生态系统的力量

  5. Demand side income

  1. 存储的数据

  Demand is by checking the data storage changes over time and directly measure,被视为DCSThe main provider ofKPI.Only from the measures,Filecoin占优势;截至撰写本文时,Filecoin存储了超过90%的DCS数据圈,而90Days before the data circle was only about82.8%.

  图 9. 存储的 DCS The proportion of data market.资料来源:Storj Stats、SiaStats、Viewblock、、Hunter Lampson.

  FilecoinNot only stores the most data,And also the fastest growth rate.在过去的90天里,FilecoinThe data stored on the platform has grown112%.

  图 10. 数据存储增长(过去 90 天).资料来源:Storj Stats、SiaStats、Viewblock、、Hunter Lampson.

  According to these are storage protocols,The amount of data stored is an important indicator,Although it has serious limitations.Data storage revenue or the data itself doesn't tell us agreement(它的价值如何,它的功能是什么,How long it will be stored,等等).在DCS和CCS供应商之间,About how to describe the characteristics of the data is stored there is a continuing discussion,Because not all the data are equal importance to(和对待).Some data is more important than the other data.Users may, according to the index to differentiate their storage provider,So the amount of data that can only describe part of.

  Data storage is also a lack of data will be how to contribute to the demand side of the agreement income background,这在考虑Filecoin时,尤其成问题,Filecoinis the only one that offers storage essentially for freeDCS服务.出于这个原因,The user may useFilecoin存储数据,Because it is the current pricing(后面会详细介绍……).Although I'm having a hard time finding open resources on this(原因很明显),但有趣的是,Countless builders and researchers in this field-They are all people I have great respect for-告诉我,FilecoinTend to cooperate with large institutions,Provide free storage,to manipulate their storage metrics.从理论上讲,FilecoinCan more than any otherDCSThe protocol stores an unlimited amount of data,but still generate zero demand-side revenue.

  2. 节点分布

  Although the data storage is the direct measurement of storage requirements,But we can also look at indirect measures.Node distribution is important to understand,Because it highlights the geographical part of demand side and supply side participants.我们可以通过观察1)Storage nodes and 2)The geographic distribution of search interests to assess this.

  Storage nodes on the space distribution of the dispersed the better,Higher dispersion(通常)Form a more decentralized,And from the node to the end user's retrieval time.High dispersion also lowered the cannot recover the risk of data loss(Is usually due to environmental factors,如自然灾害).理想情况下,Nodes are not arbitrarily scattered in space,Rather it is related to the storage requirements in the space(May be equivalent to technology of saturated multiplied by the population density of).鉴于美国、Storage demand is most concentrated in China and Europe,We expect that they will have the greatest concentration of storage nodes.因此,CCS和DCSSolution of node distribution is concentrated in the United States、Europe and China make sense.DCSNode distribution andCCSStorage center distribution similar,是DCSSolutions to important market maturity level positive signal.

  图 11. DCS The geographical distribution of node.资料来源:Filscan、Viewblock、Storj、SiaStats、Sam Williams.

图 12. CCS 节点(数据中心)The geographical distribution of.资料来源:Peterson 2015,Baxtel,谷歌.图 12. CCS 节点(数据中心)The geographical distribution of.资料来源:Peterson 2015,Baxtel,谷歌.

  3. Interested in search

  If we consider the node distribution asDCSThe distribution of supplies,那么我们可以(至少部分地)The search interest distribution asDCSThe distribution of demand.(This assumption is based on the fact that:对于DCSThe solution of each search,Searchers are more likely to be storage users,而不是供应商.)

  根据这个指标,FilecoinClearly the current worldwide with the highest interest in search of dominant,相对于Storj、Sia和Arweave.因此,人们可能会认为FilecoinThe highest relative demand.

图 13. 按国家/Relative Search Interest Advantage by Region.资料来源:谷歌趋势.注意:我使用术语“Sia”而不是“Siacoin”图 13. 按国家/Relative Search Interest Advantage by Region.资料来源:谷歌趋势.注意:我使用术语“Sia”而不是“Siacoin”

  These assumptions are based on current supply and demand indicators,但回顾过去,也可以得出类似的结论.自2021年年中以来,FilecoinConsistently the most searchedDCS解决方案.值得注意的是,2021年8月的Arweave和2021年11月和2021年12月的Storj几乎超过了FilecoinSearch interest.

图 14. Relative search interest over time.资料来源:谷歌趋势图 14. Relative search interest over time.资料来源:谷歌趋势

  Although interest in search can be a useful indicator of,But it has serious limitations.This indicator shows us a single user how to use Google to obtain information about each project,Search interest does not tell us the actual protocol requirements.

  可以很容易地得出结论,因为FilecoinRaised the most money to date,They probably have the most money to spend on marketing.那么,Search marketing budgets might explain each project interest changes.Maybe interested can search more dominant than the agreement demand forecast for money——谁又能说的准呢?此外,Filecoin拥有易于理解的、More keyword domain,如Web3.Storage and non-functional testing.存储时,may also cause data bias.用户在搜索“Web3 Storage”时可能会遇到Filecoin的服务,This is entirely based on search engine optimization (seo) and they have fields.

  Variability in search of another limitation is that it may be associated with storage requirements height not.例如,If the user plans to hundreds oftb的数据移动到DCS提供者,So they search activities(一次搜索)will not reflect their actual storage needs.It may also be an external variable,比如像Coinbase这样的cex(中心化交易所)The extent to which these individual tokens are marketed,在这里发挥了重要作用.

  4. 生态系统

  因为DCSSolutions exist at the infrastructure layer,So their ecosystems usually on behalf of the user requirements,Because their users(消费者、公司、开发人员)You can choose to use or build which ecosystem.The power of the ecosystem comes from1)Based on the agreement and the project2)With the cooperation of existing projects.Considering the maturity of their partnership and new projects add speed,FilecoinHas the strongest ecosystem.在过去的18个月里,FilecoinEcosystems have been from40Project development to300多个项目.FilecoinHas a group of impressive partners,其中包括:Chainlink、Polygon和Polygon Studios、The Graph、Near、ConsenSys、Brave、ENS、Flow、Hedera、ChainSafe、Ceramic、Livepeer、Audius、decrypt、MoNA和Skiff.

图 15. Filecoin 生态系统.资料来源:Messari图 15. Filecoin 生态系统.资料来源:Messari

  To help develop their ecosystem,FilecoinFoundation on its ecological system and funding projects invested a lot of money.Protocol Labs是Filecoin背后的团队,到目前为止已经进行了46A direct investment,向包括decrypt、Syndicate、ConsenSys和SpruceEcosystem projects including4.8亿美元.

图 16. 随着时间的推移,Protocol Labs的投资.资料来源:Crunchbase图 16. 随着时间的推移,Protocol Labs的投资.资料来源:Crunchbase

  仅次于Filecoin的是Arweave的生态系统,Filecoin有近300个合作伙伴,而Arweave大约有60个.Although many partners can benefit from the two platforms at the same time——例如,Mirror和SkiffCan provide users with at the same timeFilecoin和Arweave的服务——但其他项目,如Solana,unlikely to use both platforms at the same time.This means that many of the most criticalweb3基础设施项目——协议、dApp、NFT平台——Will find the conjunction of storage products and market agreement,And the ideological andFilecoin或Arweave保持一致,这取决于具体的用例.The strength of each ecological system will be in each platform play a crucial role in the long-term viability of the,So the ability to win the heart of the old and new builders and thought is the most important.

  值得注意的是,相对于Filecoin, ArweaveEcological system on the platform to build more new projects——Because they depend on technology to survive——rather than selectively exploiting existing technology projects.这也解释了为什么FilecoinWill work with more mature projects,这并不是因为Filecoinpartnership grows faster、更成功,而是因为Filecoin的合作伙伴(如Cloudflare和Opera)存在的时间更长.相比之下,ArweavePartners are generally in the network based on self-made early company.ArweaveSome of the notable partners includeSolana、Polkadot、The Graph、Mirror、Bundlr、Glass、KYVE、Decent Land、ArDrive.

图 17. Arweave 生态系统.资料来源:@axo_pas(在 Twitter 上)图 17. Arweave 生态系统.资料来源:@axo_pas(在 Twitter 上)

  自2020年以来,Arweave已经向15Ecosystem projects have invested nearly5500万美元,包括Mask、Fluence和Pianity.通过他们的Open Web Foundry加速项目,ArweaveTo help developers create permanentWeb应用,And through their community fund operating ecosystemARCA DAO进行了投资.

  Sia 和 StorjThe ecosystem of the small,分别有大约30个和13个项目.尽管Sia和StorjThe ecosystem of the smaller,But they have an excellent partnership.StorjSome of the partners includeCoinMarketCap、、Kraken、Filebase、Render、Akash和Quant,而Storj的合作伙伴包括Microsoft Azure、Fastly、Couchbase和Pokt.重要的是,StorjStrategy is about capturing existingAmazon S3用户建立的,Existing companies including large.因此,StorjMany of the partners may refused to publicly listed.StorjPartner might not be as the company any good.相比之下,在Arweave上构建的新的web3Native project may do benefit from is listed as a partner,To show that they in the ecological system of immersive.Different promotional motive make ecosystem more challenging,because we lack a complete dataset.

  今天,Sia主要由Filebase(第一个与Amazon s3兼容的dApp)和Arzen(A consumer decentralized storage application)使用.

  5. Demand side income

  Data storage is probably the most direct measure of the user appeal,But demand side income measure the value of user appeal——Or ability to monetize user attractiveness of the project.正如Sami和Mihai(两人都是MessariCrypto分析师)在他们关于Filecoinincome model as explained in the article,Demand side income is a useful index for infrastructure projects,Because it measures the the payment for people using the web(在这种情况下:Fees paid for storing data).重要的是,The demand income does not include the rewards paid to miners blocks.

  虽然Arweave、Sia和StorjThe demand-side revenue data can be found atWeb3 Index上找到,但FilecoinDemand-side revenue data for are hard to find(If someone can find these data,我很乐意看到).因此,我们不能将FilecoinIncorporated into the demand income compare.

  关于Filecoin的收入,我们所知道的是,Although the data stored on its platform is growing,But their income remained unchanged.This could be due to two reasons:首先,HyperDriveThe update increases the storage access rate10-25倍,resulting in lower demand for data block space(我们将在后面看到,这损害了Filecoin的代币价值).其次,In the case of do not produce higher income to store more data show thatFilecoinData is actually stored for free.因此,The miners got the reward an unsustainable blocks,For each block about20.56 FIL,This will decrease over time,在不久的将来,FilecoinWill need to improve storage of prices,To incentivize miners to participate in the network.

图 18. Filecoin Agreement income growth and ecological system still has nothing to do.资料来源:Messari图 18. Filecoin Agreement income growth and ecological system still has nothing to do.资料来源:Messari

  但是对于Storj, Sia和Arweave,We can see that in the past90Generation of demand-side revenue within the day.

图 19. Storj、Sia、Arweave The demand side of income(过去 90 天).资料来源:Web3 Index图 19. Storj、Sia、Arweave The demand side of income(过去 90 天).资料来源:Web3 Index

  因为Filecoin、Sia和StorjIs based on the contract(临时)解决方案,Where users pay for continuous storage data,So we can assume that their demand side has a nonzero part of income is a regular.相比之下,我们可以假设Arweave 100%of demand-side revenue is non-recurring,Because the user need to pay a charge can be permanently stored data.这意味着ArweaveDemand side revenue generated net is the only way to store the new data.这对Arweave来说是一个巨大的挑战,Reminds us that we willArweave与DCSCompare the disadvantages.

  May prove to be a moat——Or competitive disadvantage——的是DCSRevenue efficiency between platforms on the demand side(Roughly equal to the price)上的差异.I put the demand“Revenue efficiency”Defined as per store a byte of data generated by the demand side income.在过去的90天里,Storj和SiaEvery uploadTB分别产生了96.50美元和89.90The demand side of dollar,而ArweaveEvery uploadTB产生了约10200美元的收入.This kind of pricing model isArweave与其DCSAnother fundamental difference from competitors: ArweaveCharge a premium for services with unique features.这也意味着Arweave可以存储比DCS少113倍的数据,But still can produce andDCSCompetitors' same demand-side revenue.这表明ArweaveShould not be like other solutions to store the same number of data,Because of its services and pricing mechanism are not compare.

  Tokens valuation


  Storj, Sia, Arweave和Filecoin被理解为1)实用代币(礼品卡)和2)交换媒介(货币)的组合.Practical tokens, valuations are based on the expected utility in the future;The value of money depends on supply and demand.Practical scrip holders can use them to exchange service——在这种情况下,Is cloud storage.For a specific service for practical tokens, ability makes it structurally similar to traditional gift CARDS or vouchers.但是,Unlike gift cards and vouchers,Utility tokens are programmatically provided and autonomous,Gift CARDS and currency is in business or government to provide a,(几乎)Always in a different currency(通常是法币)发行.Supply to ensure the stylized designated supply schedule,This allows us to accurately calculate the token supply.(最近9.1%的CPIProgrammed data shows us the money supply the power)We see this combined with the traditional monetary theory,Derive the intrinsic value of each token.

  需要明确的是:I'm interested in is to use the traditional currency theory and discounted cash flow analysis to estimate tokens price changes over time.I'm not going to assess the value of the protocol itself(Generating revenue[Although income is important…]),Also do not intend to evaluated the miners or storage vendors profit.I also recognize that these protocols have special cultural value,特别是当它们(不可避免地)Became public products.也就是说,Each protocol of tokens, the price may not consider these,So I don't consider.

  首先,一个重要的区别:The traditional public securities(如$AAPL: Apple)和代币(Issued by agreement)Represent different things.Although protocols assets flow,But they do not like apple to generate cash flow.因此,Tokens should not be confused with public stocks.Tokens representing use/The right to deal;Publicly issued shares represent ownership.(Tokens can represent more than just utility/事务的权利——例如,They can also include governance rights.)随着时间的推移,Tokens of price valuation needs to include different mechanism:Monetary Theory and Discounted Cash Flow Analysis.

  Tokens of valuation models

  The main framework I use to evaluate token prices isChris BurniskeIn his seminal work:Cryptoasset Valuations(Encryption assets valuation)中提出的模型.Chris认为,With the traditional modelingDCF,It is better to keep the same structure,Replace cash flow with exchange equation,So we can deduce the current utility value of each tokens.然后,Our utility value for the future application of the discount rate to today's intrinsic value is derived.

  Replacement for the exchange equation:MV = PQHelp us incorporate the monetary nature of the token.正如Chris(And countless others)所承认的,This model has its limitations(All of the prediction model has),But it may be that we have the best models.Given the lack of a perfectly efficient market,As well as the inherent large error range to predict the future,This model is best applied to create leveraged scrip value.

  Chris写道:“The valuation of encrypted assets is mainly solved by solvingM组成,其中M = PQ / V.MIs the size of the support forPQ、速度为VThe size of the monetary base required for the cryptoeconomy of.”

  Block unicorn注释:M = Asset base size,V = 资产的速度,P = The price of the token resource provided,Q = The amount of token resources being supplied.

  Tokens valuation models:投入

  In order to forecastM, V, P和Q,I will use the following method:

  Mathematical deduction input

  1. 最大供应量

  2. 流通量

  3. Circulation compound annual growth in

  4. 存储成本(1$/GB/年)或(1$/GB)

  5. Storage costs decline in(CAGR) 

  6. The size of the data storage market

  7. The data of annual growth(CAGR)

  This is by the big three in the last decade the cost of storing an annual decline in($/GB/年)得出的,如下图所示(图20):

  图 20. AWS、Azure、Google Data storage costs and CAGR.资料来源:AWS、Azure、谷歌、Hunter Lampson.

  Subjective assumption

  1. The percentage of the hold tokens(% of uncirculated public supply)

  我假设,在2021年,50%Tokens are held.This assumption stems from the fact that:从历史上看,大约一半的CoinbaseUsers see Bitcoin strictly as an investment,While the other half as a medium of exchange.

  2. % change in tokens held each year

  I am assuming from2022年开始,The percentage of token holdings will be calculated on an annual basis1%的速度下降.随着市场趋于平衡,Less potential for value appreciation,So the number of tokens in circulation will increase(Tokens holding rate).It is difficult to estimate the—同样,It is best understood as help tokens, valuation of leveraged.

  3. 速度

  Suppose the growth rate of each token is20%,Given that Bitcoin’s speed has historically been14%左右,我在这里使用20%Is a conservative approach.

  4. TAM(Get global data market share)

  我假设Arweavecan handle the global data market10%,而Filecoin、Sia和StorjCan handle the rest90%.Permanent data storage is a whole new market,So it is difficult to determine it can handle the percentage of the existing data market,So here I am using10%,Want to keep some.临时数据存储—Today's leading storage solutions—Must account for the data of market100%.If we assume that the existing data market10%将过渡到Arweave,那么剩下的90%就留给Filecoin、Sia和Storj来处理.

  5. 获得TAM的最大百分比

  我假设TAMThe biggest percentage isArweave, Sia和Storj的1%.因此,我假设Arweavecaptures the global data market10%中的1% (ArweaveIn total, the global data market was captured0.1%,Sia和Storj分别捕获了0.9%).Given its traction and maturity,我假设Filecoin捕获了25%的可用TAM(90%的25% = 18%The global data market).

  6. 拐点

  我假设2024The year is a year each network to reach a turning point,This year has reached theTAMGet a percentage of maximum10%,这几乎是不可能预测的——another illustrative lever.

  7. 饱和/年

  I'm assuming saturated/年(网络从TAMThe largest percentage of10%到90%需要的时间)对于Arweave、Sia和Storj是10年,对于Filecoin是4年,Another impossible to predict.

  8. 折现率

  I assume that the discount rate for40%,This is the industry standard for the level of risk assets.

  Here are all the investment of fixed and variable mathematical deduction and subjective assumptions a brief view:

  图 21. Tokens valuation models used in the fixed and variable mathematical derivative inputs and the concise view of subjective assumptions.资料来源:CoinMarketCap、Uygun 和 Döngül,2021 年,Chris Burniske,Hunter Lampson.


  1. Data storage costs down(CAGR) = 0%

  2. Assuming that storage costs($/GB/年)= $0.002/GB/年

  在表中,I will clearFilecoinData storage costs down(CAGR)And assumes that the storage cost($/GB/年)marked as a very subjective assumption,Though the data is available.I do this becauseFilecoinThe current price is too low,难以持续.

  首先,Let's start with assuming storage costs($/GB/年)开始.目前,FilecoinThe storage cost is about0.0000017美元/GB/年,Or the cost of storing0.0011%On the three major providers.As I discussed above,Filecoinpricing model is unsustainable,Because it is heavily subsidized by block rewards.Since them2Since the $100 million ICO,FilecoinSubsidies on their network storage costs.As they give up subsidies,We can expect their storage costs will increase at the current level.在同等条件下,The increased cost of storage,Under the demand of fixed,使$FIL更有价值(Like any tokens),但我们可以假设,随着Filecoininevitably raise prices,Storage requirements on its network may be reduced,降低$FIL的内在价值.

  It's hard to say how will the team performed to raise prices,Even if the price is still lower than the big three.If we take the current price of about0.0000017美元/GB/In running the model,则2022In the intrinsic value is about0.00美元/FIL,再次表明FILToday's pricing model is unsustainable.因此,我估计未来10年FilecoinStorage costs as0.002美元/GB/年(Cheaper than three storage price100倍)(Assume that data storage costs down[CAGR]为0%).这保持了FilecoinThe price competitiveness of——make them cheaper than the big three solutions100倍——At the same time as tokens price offers significant value.View this investment as a personalFilecoinExpectations for the sustainable development of,Even the requirement.

  Tokens valuation models:投入

图 22.Tokens valuation models into.资料来源:CoinMarketCap、Hunter Lampson图 22.Tokens valuation models into.资料来源:CoinMarketCap、Hunter Lampson

  在同等条件下,Model in leverage as follows:

图 23. Illustration of various levers that generate token value.资料来源:Hunter Lampson图 23. Illustration of various levers that generate token value.资料来源:Hunter Lampson

  This table is given each lever changes generally put,而不是保证,When the lever to increase/Reduce the arrival ego/Low Numbers,Input is always the right.例如,考虑速度:平均而言,随着速度的增加,代币价格下降.But the speed of an arbitrary low levels,For example, at the rate of0,Will mean the token is traded annually0次,因此需要0monetary base to satisfy the ecosystem.也就是说,By avoiding the ending,The general trends cited in the table are useful.


  Arweaveeconomic benefits are the most defensive,Its driver is relatively low$ARTokens, supply and storage costs are relatively high($/GB).This part builds on my previous conclusions,即Arweave是DCSThe product with the highest revenue efficiency on the demand side,This means that it can be stored less than similar products113倍的数据,and generate the same demand-side revenue.此外,我假设ArweaveTo capture the global data circle0.10%,This assumes that the conservative enough,是合理的.如果实现这一目标,2032Tokens, prices are expected to increase from current levels+182.91x.尽管ArweaveThe higher relative pricing may strengthen its unit economic benefit,But it may also be a hinder the user USES the Achilles' heel,Users will eventually decided toArweaveThe service is worth.

  Even if we assume that users are willing to pay for the additional cost,理论上,They must be persuaded to use the product in practice.因为ArweaveProducts with competitors have different nature,Switching costs may be too high,The service also very unique,Can't win new users.尽管Arweave有潜在的优势,But the high cost and the dependence on the new market may become insurmountable obstacle.如前所述,ArweaveThe only way to produce the demand side income is to store the new data.从表面上看,ArweaveDoes not seem to have every bit of data demand side benefits——这是所有CCS和DCSof competitors can benefit.相反,我认为ArweaveBenefit from demand-side benefits of involuntary duplication.ArweaveNot to the user charge permanently,But in advance“Permanent recurring demand-side revenue”,这可能是ArweaveOne of the most valuable donation mechanisms.

  目前,Because of its low price,FilecoinEconomy is the most unreliable.Given a fixed token supply,Lower costs for utilities,The monetary base to support it has to be smaller.This view will be priced lower value is defined as tokens of negative attributes,Rather than a positive attribute.同样有可能的是,FilecoinThe low pricing is widely used for its laid a solid foundation.Low pricing can also beFilecoin的关键区别,This may be a necessary moat.

  然而,My concern is the pricing power is decidingFilecoinwill play an important role in the future.正如Tushar和Spencer所说,Filecoin(以及Sia和Storj)Are directly compete with the big three temporary storage market.With the big three on price could be disastrous.如果FilecoinCan be in the absence of unsustainable subsidies keep low price,Then its maturity、The strength of the ecological system and the influence of the whole industry will make it a big three of the most powerful challenger.If it turns into a price war——这可能是不可避免的——things could get bad.

  根据该模型,Based on the current pricing and2032Difference between yearly price forecasts,SiaThe token economics of making it worth more thanStorj高出4.5倍.通常情况下,Sia和Storj被归为Filecoin的弟弟.鉴于Sia和/或Storjof less robust ecosystems,Hard to imagine in the near future,They will replaceFilecoinDominance in this field.尽管如此,Sia和StorjScrip economy thanFilecoinScrip economy more attractive.Pricing for tokens, valuation and the long-term viability of each project are indispensable.

  Limitations and Reflections on Future Research

  1. 云存储≠云计算.正如Christine Deakers所指出的,Many cloud storage users at the same time for their stored data using cloud computing.DCSThe solution must solve the problem.Filecoinhas started building its virtual machine——其他DCSSolutions may follow suit.

  2. DCSSolutions need more integration.正如Mark Gritter所指出的,Most of the Internet of things application needs not only distributed storage,Decentralized database is also required.如果DCSSolution without the machine integration with the traditional time series database,This may be the main obstacle.

  3. DCSThe solution should allow for location selectivity.Mark GritterOne example mentioned is self-driving cars.Self-driving cars collected sensor data streams must be stored in decentralized way,to achieve the lowest possible latency.If the data uploading(Cars and motors)Unable to select a location near the store data,DCSSolutions may prove to be a good solution to this case.


  (1)Although cloud computing is different from cloud storage,But we can reasonably make a set of assumptions:首先,To provide cloud computing services(Such as the big three)Tend to provide this service in their stored data.换句话说:Customers usually on a platform at the same time the use of computing and storage services.

  第二:我们可以假设,As cloud computing company occupy more market share,Them from the better unit economic benefit from the faster and faster.The bigger a company is,It's on the hardware pricing negotiation more effective,This will reduce the cost of customer,吸引更多的用户,To further increase their bargaining power.所以,When I mentioned that the Big Three occupied65%The cloud computing market share,We can assume that they occupy a similar number of cloud storage market share.

  (2)在这篇文章中,我使用术语“安全”和“安全”To describe the highly replicated data on distributed node set,This leads to higher data redundancy,More consistent uptime,And reduce the possibility of a review and a single point of failure risk.


copyright notice
author[Sina Finance-Blockchain],Please bring the original link to reprint, thank you.

Random recommended