​Decentralized cloud computing track research: breakdown of AWS competitors

As the application of Web3 becomes more and more abundant, the performance requirements for data computing are also getting higher and higher. In response to the computing needs of more data assets, the decentralized cloud computing track is also booming. This article attempts to sort out this track and analyze the main projects Filecoin, Arweave, ICP, and the new projects Ceramic and Fluence.

This article directory:

1. Introduction to the track

2. Filecoin: Storage Pathfinder Based on IPFS Protocol

2.1 Core Mechanism

2.2 Token Economy

2.3 Operational Status

2.4 Team and financing

3. Arweave: Web 3.0 full-stack protocol featuring persistent storage

3.1 Core Mechanism

3.2 Token Economy

3.3 Operational Status/Team and Financing

4. ICP: World Computer Public Chain

4.1 Core Mechanism

4.2 Token Economy

4.3 Operational Status

4.4 Team and Financing

5. Ceramic: Middleware in the field of database storage

5.1 Core Mechanism

5.2 Operation Status

5.3 Team and Financing

6. Fluence: p2p development platform and decentralized hosting market

6.1 Core Mechanism

6.2 Operational Status

6.3 Team and Financing

7. Summary

1. Introduction to the track

Cloud computing is a huge market. According to the definition of the National Institute of Standards and Technology (NIST), cloud computing “is a model that can realize anytime, anywhere, convenient, and on-demand access from a shared pool of configurable computing resources. Obtain the required resources (such as networks, servers, storage, applications, and services), and resources can be provisioned and released quickly, minimizing the workload of managing resources and interactions with service providers.” Cloud computing is a broad concept that includes all IT infrastructure resources that can be accessed on demand.

End-user spending on public cloud services worldwide will reach $396 billion in 2021. Among them, Amazon AWS often tops the list with a market share of about 30%, and other mainstream cloud vendors are also companies that provide centralized services.


Global end-user spending on public cloud services, source: Statista


Market share of cloud infrastructure service vendors, source: Statista

Web3 broke out a series of landing applications in the 2020 Defi Summer, and the core value of these applications lies in the confirmation of ownership, more precisely, the ownership of financial assets. Financial assets exist in the form of blockchain ledgers, which are essentially cloud services. However, the scale of data in the financial field is small, and users can still afford mainstream public chains even with limited performance and high fees. With the development of Web3, landing applications have begun to develop from the financial field to image art, games, social networking, etc., and ownership has also expanded to richer digital assets. The Web3 world has an increasing demand for data computing and storage.

However, the original mainstream public chain with small capacity is difficult to carry, and centralized cloud services cannot meet the needs of Web3 users for security, openness, and anti-censorship. At present, most Dapps only realize the decentralization of the logic layer, and the storage and calculation of a large amount of data are still running off-chain. In the field of Web3, the demand for cloud computing is also expanding.

The expansion of the public chain and cloud services in various vertical fields by the Web3 project are essentially responding to such needs. With the mainstream products of AWS, we can also understand the development picture of the current decentralized cloud computing market.


AWS’s main products and benchmarking projects in the decentralized field

AWS’s main product suite consists of three parts, which also represent several mainstream areas of the cloud computing market.

One is S3 (Simple Storage Service) for storing large or static data. In this field, its decentralized benchmarking projects include Filecoin and Arweave based on IPFS.

The second is a database for specialized data storage, including relational database RDS and non-relational database Dynamo DB. Decentralized benchmarking projects in this field include Ceramic and so on.

The third is EC2/Lambda for cloud computing and hosting. EC2 is more flexible but more complex to configure, while Lambda is less flexible and easier to configure. In this field, decentralized benchmarking projects include Fluence, which is similar to Lambda for cloud computing, Livepeer, a video transcoding platform, and Render Network, a graphics rendering platform. In actual work, these tools will interact with storage objects such as S3 and RDS respectively, and then return to end users accessing AWS.

Finally, AWS as an overall product suite also has its on-chain competitor: Dfinity. Dfinity is committed to becoming a world computer, providing solutions for all IT infrastructure such as computing and storage on the chain. If Dfinity really achieves its goal, it will become the end of the public chain, with a grand vision and a long way to go.

Below, we will analyze the projects of Filecoin, Arweave, ICP, Ceramic, and Fluence respectively.

2. Filecoin: Storage Pathfinder Based on IPFS Protocol

Filecoin is an incentive protocol based on the underlying storage network of IPFS. The two were founded successively by Protocol Labs. Filecoin’s current FDV is about $40 billion.

IPFS (The InterPlanetary File System) is a network transmission protocol for distributed storage and sharing of files. It was founded in 2014 and has an open beta version in 15 years.

In 2017, Potocol Labs issued Filecoin, an incentive layer protocol based on IPFS, and completed a huge public offering of $257 million on CoinList. Filecoin supports users to pay on demand, similar to AWS’s S3 service model.

2.1 Core Mechanism

Merkle DAG for IPFS: A data structure that facilitates file query and download

IPFS adopts the data structure of Merkle DAG, which is a transformation based on Merkle Tree. Through this data structure, IPFS implements content addressing and fragment downloading of files.

IPFS assigns each file a unique hash, similar to a file fingerprint. Each root file will point to multiple node files. Once the content of the node file changes, the hash value will also change accordingly, so that the hash of the root file will also change.


Source: IPFS Draft3

When querying files, the IPFS network searches based on the unique file hash value in the entire network. This method is also called “content addressing”. In contrast, HTTP queries files by URL address. Once the file in the address is changed, the visitor cannot find the original file. Moreover, content addressing also ensures that IPFS will not be stored repeatedly, saving network space.

When downloading a file, you only need to know a hash address, and you do not need to download the entire file. On the other hand, BitTorrent must use a torrent file when downloading, and put all the addresses of the downloaded content in this torrent file, which is not conducive to the download of fragmented files.

But content addressing also has the corresponding drawback, that is, once the file changes, a new hash value is generated. This is beneficial for file version preservation when major changes occur, but it is not suitable for content such as databases that require daily updates. A complete snapshot is saved for each update, resulting in a large amount of data redundancy. Ceramic mentioned later is committed to solving this problem.

Filecoin Consensus Algorithm: Constantly Checking the Proof of Miners

Filecoin is divided into storage market and reading market, and users need to form transactions with miners respectively.

After forming a storage transaction order, miners will store data in sectors (Filecoin’s basic storage unit, similar to a container for storing data), and prove their storage to the Filecoin blockchain. First, miners need to encapsulate sectors through proof of replication (PoRep), which proves that user data is stored, as the initialization of storage. After that, the protocol will continuously check whether the miners have saved the data, and the miners need to generate a proof of space-time (PoSt) on the sector to prove the continuous storage, and obtain block rewards and storage fees. After the sector expires, it will be terminated naturally. If the miner defaults or fails for too long (14 days), the protocol will actively terminate the transaction, return the user’s remaining order fee, and punish the miner for paying this fee.


Sector life cycle representation, source: “Creating the Filecoin Economy”

In the reading market, users put forward specific query requirements, and the retrieval miner nodes broadcast the query. The retrieval miners with this content will return retrieval proposals containing price information to trade with users.

The design of Filecoin does realize the verification of distributed storage, but the complexity of the system is very high, and it is necessary to continuously challenge storage miners to verify continuous storage, resulting in high storage costs; moreover, miners need to pledge tokens before completing the order, which is relatively High implied option costs.

2.2 Token Economy

The total amount of FIL tokens is capped at 2 billion. The initial distribution is shown in the figure below. 55% will be released slowly as storage mining rewards, 15% will be used as mining reserves, and will be used by community proposals in the future. 10.5% will belong to the Protocol Labs team, and the rest Attributable to early investors, Filecoin Foundation, etc.


FIL token distribution chart, source: “Creating the Filecoin Economy”

The reward for storage mining consists of two parts, 30% can be obtained as long as the order contract is met, and 70% can be obtained only when the entire network reaches the baseline of computing power, otherwise the reward will be delayed to encourage more miners to participate in storage. The baseline is initially set at 1EiB (less than 0.01% of global storage) and grows at a rate of 200% per year. When storage reaches 1~10% of global storage, the community can decide to slow down the growth of the baseline.

If the sector is terminated early, the FIL pledged by miners will be confiscated and destroyed as a punishment, so FIL has a certain deflation mechanism.

In addition, Filecoin also verifies and confirms the user’s real demand for stored data through a globally distributed physical network. After verification, the storage reward of user data is higher, preventing miners from earning rewards by storing junk data. However, this “loophole-fixing” design has also been criticized a lot, because it is difficult to be completely objective about what is the confirmation standard of real data, and the adjustment to adapt to various needs will become more and more complicated.

2.3 Operational Status

Filecoin’s current total storage is 15.52EiB, or about 16 million TB. Compared with Arweave’s only 50TB, Filecoin can be described as huge.


However, combined with its mining reward data, Filecoin’s economic model seems to have a lot of hidden dangers.

Filecoin’s current storage price is actually very cheap, only $0.000001/GB/month. Filecoin compares the price of Amazon S3 to store infrequently accessed data, the former is only 0.01% of the latter.


Compared with the total storage capacity of ~15.52EiB, according to the above price, the total revenue contributed by all storage users in one month is 15.52*1024^3*0.000001≈167,000 USD; while the reward in the past 24H is about 300,000 FIL, which is estimated to be issued in the past month The total reward of about 180 million US dollars, it can be seen that the actual fee paid by the user is much smaller than the mining reward.

No FIL miners were interviewed at the time of writing, and the true storage cost is uncertain. However, it can be seen from the above figures that the network as a whole should be supported by token mining, and a healthy business model has not been formed. 55% of the total 2 billion tokens, that is, 1.1 billion FIL will be used as mining rewards. At present, a total of 160 million pieces have been issued, minus the 30 million pieces destroyed, a net increase of about 130 million pieces, and there is still room for 970 million pieces to be issued. However, if the currency price falls in the future, there may be a risk of miners withdrawing and losing data.

2.4 Team and financing

Protocol Labs, the parent company of Filecoin and IPFS, was first founded in 2014. According to the information collected by Crunchbase, Protocol Labs has raised a total of 5 rounds of financing, with a total financing of 10.9 million US dollars; Filecoin has completed 7 rounds of financing, with a cumulative financing of 258 million US dollars. The above financing includes ICO, and investment institutions include a16z, Sequioa Capital, BlueYard Capital, etc.

Founder Juan Benet, a Stanford CS major, started his business in 2010, and has worked on mobile games and media mixers.

3. Arweave: Web 3.0 full-stack protocol featuring persistent storage

Arweave is the largest old-fashioned storage protocol other than IPFS. It was launched on the main network in June 2018, and the current FDV is about 2 billion US dollars.

Arweave itself includes a storage layer and an incentive layer, and is a complete decentralized storage protocol, equivalent to IPFS+Filecoin. In terms of storage function, Arweave features a one-time payment and permanent storage.

At the same time, Arweave also includes the Smartweave smart contract protocol, which has the potential to realize the calculation and storage of smart contracts at low cost and solve the scalability problem of smart contracts. In this sense, Arweave goes beyond storage and is a Web3.0 full-stack protocol.

3.1 Core Mechanism

Regarding the specific implementation mechanism of Arweave, there are already many detailed Chinese materials, please refer to the appendix. This paper selects the core mechanisms of storage proof, Permaweb, and Smartweave smart contract protocol as background introduction.

Proof of Storage: Low-Cost, High-Reliability Persistent Storage Networks

Filecoin proves storage by checking the execution of each contract. This mechanism leads to high proof costs and is difficult to match the performance of centralized storage. Arweave’s solution to proof of storage is to randomly check, and miners who store randomly determined copies will be rewarded, which greatly reduces the cost of proof and encourages miners to store data with sparse copies.

Arweave builds a blockchain-like data structure called Blockweave, where each block is connected to a previous block and a memory block in the history, and the memory block is randomly determined. Arweave uses Proof of Access PoA, and only miners who have stored memory blocks have the opportunity to participate in PoW guessing. Miners can decide which blocks to store by themselves, and do not need to store all blocks, but since the memory blocks are randomly determined, in order to be able to access the memory blocks to get rewards, miners need to store as many blocks as possible, or in the case of limited space Store rare blocks to improve guessing odds. To put it simply, Arweave encourages storage through the design of “the more storage blocks are rarer, the greater the reward probability”.

Compared with Filecoin’s independent inspection of the design of each contract, Arweave greatly reduces the cost of proof, which also reduces the cost of storage; and according to calculations, the probability of Arweave losing data is extremely low, the order of magnitude and the probability of collision between two private keys are the same, and the reliability Very high.

In addition, Arweave also uses the Wildfire node scoring mechanism to encourage nodes to respond to reading; the Blockshadow block shadow only verifies the design of the hash value to improve the efficiency of block verification and invocation.

The current cost of storing 1GB of data in Arweave is about 0.25AR, or about $7.89.


Permaweb is always on the web: Permanently retain the application layer of the Dapp front end

Permaweb is an application layer based on the Arweave network protocol. Dapp developers can deploy the front-end through the Permaweb network, but the front-end content is packaged and stored in the Arweave network, realizing the permanence and censorship resistance of front-end access. . Since the content that is always on the Internet cannot be changed, after the new version is updated, multiple versions will be left.

Smartweave: a smart contract protocol that extends the logic layer of Dapp

Smartweave is Arweave’s smart contract protocol, which allows developers to develop smart contract Dapps with Javascript. The contract code and contract creation status are stored in the Arweave network. The difference from public chains such as Ethereum is that usually the public chain requires all full nodes to perform the verification and calculation of all data on the chain, and the redundancy is extremely high, but Smartweave only needs a single node of the contract caller to verify, from the contract The genesis starts to execute, and finally submits its own transaction data and the contract status after executing the transaction to the Arweave network.

Smartweave reduces the burden of network computing and may become an important way to improve the scalability of the Dapp logic layer. In this sense, Arweave is not only a storage protocol, but also a Web3.0 full-stack protocol that includes storage and smart contract ecology.

3.2 Token Economy

The primary role of AR tokens is to pay storage fees. Arweave’s model is one-time payment, permanent storage, and free reading. Of the AR tokens that users pay each time, about 14% are given to miners as instant incentives, and 86% are entered into the donation pool. Funds will be called from the donation pool only when the miner’s immediate incentive is not enough.

The current price of storing 1MB of data on Arweave is 0.6~0.8 cents, while on AWS or Alibaba Cloud, it costs at least 2 cents to store 1MB of data for 100 years. That is to say, even without considering the future cost reduction, according to the current cost and AR currency price, the data storage on Arweave can be guaranteed for at least 300 years. According to the official website, the cost of one-time storage guarantees at least 200 years of storage, which should be speculated based on the early currency price.

AR tokens, like Bitcoin, use a deflationary model and have a cap on the total amount. 55 million ARs were generated in the genesis block, and additional ARs were issued in each block thereafter. However, AR’s additional issuance decays faster. Bitcoin is halved every four years, while AR is halved every year. The total amount of AR is capped at 66 million. Currently, ~64.6 million AR has been generated, and the pending mining rate is less than 3%, which is already lower than 10% of Bitcoin. The full distribution of AR is as follows:


Among them, 13% is allocated to the project team and 26.5% is allocated to future projects, both of which have a 5-year unlock period, with 20% unlocked each year.

3.3 Operational Status

In the past year, benefiting from the development of NFT and creator economy, Arweave has made rapid progress in both storage scale and ecological development.

Rapid growth of storage scale

Since 2020, Arweave has been providing storage services for public chains and well-known Defi projects. Solana and Nervos all store the data layer in Arweave, and Synthetix, Uniswap, Curve, etc. all store the front-end presentation layer in Arweave. In addition, Arweave also provides block data storage services for Avalanche, Celo, Near, Cosoms and other public chains through the storage middleware project KYVE.

Since September last year, thanks to the explosion of NFTs and the creator economy, Arweave’s storage scale has also grown rapidly. Well-known artists Beeple and Jay-Z have launched their own works based on Arweave. Arweave’s current data storage capacity is about 50TB, compared with only 4.6TB in the same period last year, a year-on-year increase of ~10 times.


Source: https://viewblock.io/

Arweave’s 30-day direct network fee is about $288,000, which ranks first among the Web3 basic service agreements (Filecoin seems to be out of this list due to data accuracy issues, and cannot be compared with Arweave).


Ecology is flourishing

In addition to directly storing data in Arweave, there are a number of projects based on Arweave to expand services, including Bundlr Network for bulk uploading of packaged transactions, ArConnect as Arweave identity and wallet, ArDrive as Arweave storage front end, as a cross-chain KYVE, which stores middleware, etc., has gradually formed a rich ecology.


Source: Verto Twitter

The number of Dapps developed using Smartweave in 2021 has also grown significantly. Many projects in the above ecosystem are developed using Smartweave, including: Verto, RedStone, Pianity, Koii, Arweave News, etc.


Source: Arweave News

3.4 Team and financing

CEO Sam Williams, German, studied at the University of Nottingham, UK from 2011 to 2014, and received a bachelor’s degree in computer science. After studying for a Ph.D. at Kent University, he dropped out. From 2014 to 2017, Sam taught briefly at Kent University and co-founded Minimum Spanning Technologies and Arweave in 2017.

Since its establishment, Arweave has raised a total of 22 million US dollars. The investment institutions include a16z, Multicoin, Coinbase Ventures and other leading institutions.

4. ICP: World Computer Public Chain

The full name of ICP is Internet Computer Protocol. It is a public chain project led by the Dfinity Foundation. The vision is to build a public chain with unlimited capacity, so that all smart contracts can run with the performance of Web2 and become a world computer platform. ICP’s vision is extremely ambitious. The solution includes computing and storage, providing complete cloud services for smart contracts, and benchmarking AWS’s large and comprehensive product suite.

4.1 Core Mechanism

The design system of ICP is relatively large. This article selects containers, consensus algorithms, and NNS (neural network systems) for a brief introduction.

Containers: Enable efficient interaction between smart contracts

ICP consists of four layers of components from bottom to top: 1) the bottommost data center (Data centers), distributed around the world; 2) hardware nodes (Nodes) running in the data center; 3) independent data centers in the Nodes are combined into subnets. The combination of nodes is randomly determined to ensure that the subnet cannot be destroyed. A subnet can be regarded as a chain; 4) There are multiple containers (Canisters) running in the subnet, and each container is similar to a smart contract. , which is essentially WebAssembly, and all languages ​​that can be compiled into wasm can write ICP contracts. At present, the Rust SDK is relatively mature, and ICP also provides the native contract development language Mokoto.


Containers can interact with each other, providing a basis for mutual calls between smart contracts. A smart contract can call a third-party contract as a service to build ICP’s Lego.

The smart contract interaction provided by ICP is more efficient. The mainstream applications of Ethereum can also call each other, and the feature of Ethereum calling is guaranteed by the blockchain, which requires a global lock, which has high security but low efficiency; ICP is based on the actor model call, sacrificing a certain amount of Security, but contracts can be parallelized, which improves execution efficiency.

However, the interaction between containers is not perfect. First, ICP does not naturally guarantee atomicity like Ethereum, so there is very little actual communication between different containers at present. There are already projects in the ecosystem trying to develop middleware services to achieve the atomicity of contracts. Second, cross-container communication requires at least 2 rounds of consensus, plus network delay, the delay experienced by the user will reach 6~8 seconds, which is also a hindrance for the project.

The interaction between containers has another potential use, which is to enable the scaling of a single smart contract. The current memory limit of a single container is 12GB. If a project requires too much memory, multiple containers may be required to host it. But due to the above problems, this is not a perfect choice. In addition, ICP is also exploring the memory expansion of a single container. If it can be achieved, the need for expansion through cross-container interaction may disappear.

Chain Key: The core technology to realize expansion and cross-chain

There is an underlying technical solution in ICP’s consensus algorithm: Chaink Key, which uses BLS threshold signature (also known as TBLS) to provide consensus services for ICP. TBLS can realize signature aggregation. Nodes that reach a certain threshold will create a group signature after they independently sign. Any subset of groups that reach the threshold can generate the same signature, which can be verified with the public key of this group.

This solution enables unlimited new subnets to be added to the entire network, which forms the basis for the unlimited expansion of ICP. In the actual verification process of each block, ICP will randomly select a set of threshold groups from all nodes. After the number of signed nodes in the threshold group reaches a certain threshold, a group signature will be formed, as long as the public signature of this group is used key can be verified. This is the power of Chain Key: no matter how many nodes there are, security verification can be achieved through a randomly selected threshold group, and it is no longer necessary to verify all nodes.

Moreover, this solution is also the basis for ICP to realize communication between subnets, that is, between chains. When subnet A needs to send a message to subnet B, a consensus signature is reached in subnet A and sent to subnet B, which can be verified as long as subnet B has the public key of subnet A. The interaction between subnets, that is, between chains, can be carried out directly, unlike Polkadot, which relies on the relay chain.

NNS: realize automatic management of projects

NNS (Neural Network System) is a special container contract and a very important part of the overall architecture.

As an automated ICP management protocol, NNS is responsible for approving new node joining, IC subnet generation and reorganization, ICP protocol upgrade, etc. By staking ICP, users can participate in NNS voting.

NNS also plays an important role in ICP’s token economic model. NNS will generate new ICP tokens to reward the behavior of running nodes and voting; on the other hand, the owner or manager of the container needs to consume cycles (aka gas) to ensure the operation of the container, and cycles can be regarded as a stable currency , ICP tokens need to be exchanged one-way (ICP can be exchanged for Cycle, Cycle cannot be exchanged for ICP), and NNS adjusts the exchange rate between ICP tokens and Cycle by approving oracle proposals. It can be said that NNS is similar to the central bank in the ICP economic system, and has the function of issuing currency and managing exchange rates.

It is based on NNS that ICP has an open, decentralized and automatically implemented management method, which can be said to be a complete DAO.

4.2 Token Economy

ICP token, also known as ICPT, has two core functions: 1) Pledge in NNS to vote, the longer the lock-up period, the greater the voting weight; 2) ICP is exchanged for cycles to recharge the container operation, cycles anchor XDR is determined, the latter is a stable currency obtained by a weighted summation of a series of legal currencies, so the price of cycles is basically stable. ICP corrects the exchange rate between ICP tokens and cycles in real time by validating an oracle’s price proposal.


ICPT has both inflation and deflation models. On the one hand, ICPT is additionally issued to reward voters and node operators, and on the other hand, ICPT is consumed to run containers. The smaller the amount of lock-up and the more bonus issuance, the more ICPT will be flooded; the more the amount of lock-up and the greater the project’s demand for containers, the more scarce the ICPT will be. The price and quantity of ICPT depend entirely on supply and demand and are difficult to predict.

The current storage price on ICP is about $0.47/GB/month, which is much higher than AWS about $0.02/GB/month.

The initial token distribution of ICPT is shown in the figure below. The current total supply is about 480 million pieces, and the FDV is about 9.6 billion US dollars.


List of token distribution, source reference appendix

4.3 Operational Status

ICP has launched the main network and issued tokens since May 21, but the current ecology is not prosperous enough. By the end of 2021, the total transaction volume of ICP was 1.8 million, while there were more than 1 million transactions per day on Ethereum in the past year, with a huge disparity in volume.


There are not many smart contracts based on ICP. As of 21 years, there have been 16,000 smart contracts. Most of the project types are infrastructures such as wallets and browsers, or social applications similar to mainstream Web2 applications. Comparing with Reddit’s DSCVR, instant messaging software Openchat, etc. However, the growth of social applications depends on the network effect formed by persistent traffic accumulation. In the short term, only relying on the characteristics of decentralization will inevitably fail to compete with mature Web2 projects.

Relatively, there are few Defi and Gamefi projects native to Web3 in the ICP ecosystem. This may be due to the lack of EVM compatibility, existing mature projects cannot be copied and pasted, and ICP cannot inherit the existing liquidity in the encryption world; however, the architecture of ICP is very independent, and the application of its technical highlights needs to wait for a new project. In addition, ICP does not have a unified token standard, which also hinders its ecological development.

However, in 2022, ICP plans to integrate BTC and ETH to accelerate the development of Defi assets. After the integration of BTC and ETH, the smart contract in each container will have the function of BTC/ETH wallet, and the cross-chain transfer of assets will no longer require a cross-chain bridge, which will open up a new way of asset cross-chain. It is conceivable that this will become an important bargaining chip for ICP to compete with other public chains.

4.4 Team and Financing

The Dfinity Foundation is a non-profit organization founded in 2016 as a project incubated by String Labs. Headquartered in Switzerland, Dfinity has a large team with a total of more than 200 people, and has a luxurious technical team, including top scientists from the blockchain industry and senior engineers from Google, IBM and other companies.

Dfinity received $61 million and $102 million in financing in February and August 2018, led by a16z and Polychain. Dfinity should have raised an earlier round of ~$30 million, but it has not been announced, and it has raised about $195 million in total.

The founder and chief scientist Dominic Williams is a cryptographer and entrepreneur, as well as the co-creator and CTO of String Lab. He has made MMO games based on self-built distributed systems.

5. Ceramic: Middleware in the field of database storage

Ceramic is a decentralized network that focuses on storing high-frequency update data. Based on the IPFS storage network, it has built a new model that facilitates data update, and its function is similar to a decentralized database. The test network was launched in January 21, and the main network was launched in June 21. Several products have used Ceramic as middleware, but Ceramic has not released a token-based business model.

5.1 Core Mechanism

Ceramic’s main market is database storage. Although it belongs to the category of cloud storage with Filecoin/Arweave, the usage scenarios are different.

Filecoin implements content addressing by generating a unique hash value for each file. Once the content changes, the hash value will change, which is equivalent to generating a new file; Arweave focuses on permanent storage, and the content cannot be changed after it is on the chain. Unless another new content is uploaded. This mode is no problem for storing static files such as music, pictures, videos, etc., but it is not suitable for data that needs to be updated continuously, such as user transaction orders, user attention relationships, etc. Once updated, a new snapshot document needs to be generated. snapshots will cause data redundancy.

In the mature solutions of Web2, the storage of large static files and database storage also correspond to different storage designs and corresponding products. For example, in the product suite of AWS, Amazon S3 focuses on scenarios such as large file storage or long-term storage. Amazon RDS/Dynamo DB focuses on database scenarios. In Web3, Filecoin benchmarks Amazon S3, and Ceramic benchmarks Amazon’s non-relational database Dynamo DB.

Ceramic’s storage design has some of its own definitions, which are not easy to understand at first glance, but it is easy to understand with the logic of the database. The gray block combination in the figure below is a “Stream”, which is similar to all the historical data of an object. The initial state (Genesis Commit) and each subsequent change (Commit) are records on IPFS. These records are combined It becomes a Stream. Since the Stream records “changes” rather than “snapshots” of the resulting state, it is necessary to process all events on the Stream to obtain the latest record of the object.


Taking the database as an example, the Ceramic record mode is: initially, Alice and Bob each have 10 yuan; the next day, Alice transfers 5 yuan to Bob; the third day, Bob transfers 3 yuan to Alice. This is also very similar to the blockchain ledger, the ledger does not write the balance of each user, and all intermediate processes need to be calculated to obtain the final user balance.

In contrast, the traditional recording mode of IPFS is: file a, Alice and Bob have 10 yuan respectively; file b, Alice has 5 yuan, Bob has 15 yuan; file c, Alice has 8 yuan, Bob has 12 yuan. Here, each record is a snapshot of the resulting state, and a new snapshot needs to be generated whenever there is a change.

Ceramic ensures that each object has a unique Stream ID through such a design. With a globally unified name, the name will not be changed due to changes in content. Each write requires user authorization, and the entire process is similar to blockchain accounting, except that what is written is not transaction data, but other data, such as user account information.

5.2 Operation Status

Ceramic has begun to provide services for many Web3 projects, including RabbitHole, BoardRoom, GeoWeb, etc. mentioned on the official website, as well as projects such as Cyberconnect, which use Ceramic to store frequently updated data such as user identity information and variable NFT data. The more Web3 social projects develop, the stronger the demand for Ceramic will be.


Project case using Ceramic

The project as a whole is still in its infancy. At present, the product can only be used by developers, and there is no token economic model as an incentive layer. Therefore, the project party needs to run nodes by itself, or trust third-party nodes on IPFS, which is not a complete decentralization. ization agreement.

The future development of Ceramic, on the one hand, depends on the development of the needs of the Web3 social project itself, on the other hand, on the design of its incentive layer.

5.3 Team and Financing

Ceramic’s development team is 3Box, which was founded in 2018. CEO and co-creator Michael Sena worked as a product manager at Consensys.

3Box received $2.5 million in funding led by PlaceHolder in June 2019, and $30 million in February this year led by Multicoin and Union Square Ventures, for a total of $32.5 million.

6. Fluence: p2p development platform and decentralized hosting market

Fluence is a p2p application development platform for developing and running applications without centralized cloud services, and its positioning is similar to AWS Lambda.

6.1 Core Mechanism

The core development tool of Fluence is an open source programming language Aqua, which allows developers to easily develop p2p applications to achieve different degrees of decentralization: running software through a limited set of servers, or directly connecting user devices to achieve complete p2p communication.

Fluence also plans to build a hosting marketplace: users can choose which nodes run the software, and pay the nodes directly, with developers earning a portion of the revenue.


Based on the p2p development tools provided by Aqua and the decentralized hosting market, Fluence has conceived a series of application scenarios, including the development of social software for p2p communication, the application of nodes running by DAO itself, and the acquisition of data through IPFS and other decentralized data sources and complete the calculations and so on.

6.2 Operational Status

The application of Fluence is still very early, and it does not provide a complete incentive layer scheme, nor does it have corresponding tokens. However, if the hosting market is to be realized in the future, the design of the incentive layer must also be an important part.

At present, there are only 13 running nodes displayed on the official website, and no complete application cases are displayed.


6.3 Team and Financing

The Fluence team was established in 2017, and Tom Trowbridge, one of the co-creators, was the chairman of Hedera Hashgraph.

Fluence received $9 million in financing led by Multicoin in February this year. The investors also include Alameda Ventures, Tiger Global, Protocol Labs, Arweave Capital and other institutions.

7. Summary

Finally, we try to summarize the application scenarios and security levels of the above projects.

In terms of application scenarios, it can be seen that the entire track is transitioning from general-purpose to vertical-specific. A few relatively early projects have deployed general-purpose storage and even a larger public chain market.

Protocol Labs started IPFS in 2014, and then used Filecoin as an incentive layer to form a complete storage protocol. The flexibility of on-demand storage, low storage prices, and early token incentives make it the largest general-purpose decentralized storage project.

Dfinity, founded in 2016, has a grand vision of becoming a world computer, and universal storage is only part of its vision, which will include all software deployment and computing in addition to storage. Even in the public chain market, it can be said to be the most versatile project.

Arweave, established in 2017, focuses on permanent storage, which is more suitable for storing static cold data in terms of scenarios. However, Arweave’s ambitions do not stop there. As a Web3 full-stack protocol, it also provides the Smartweave smart contract protocol, hoping to carry the front and back ends of the Dapp at the same time.

Projects that have been active recently are more focused on the vertical market and provide services as middleware. 3Box, established in 2018, is also the development team of Ceramic. Currently, it focuses on developing database storage and provides database services based on existing public storage protocols such as IPFS/Arweave. Fluence, established in 2017, focuses on providing p2p development tools, building a decentralized hosting market, and providing cloud computing services on existing public data sources. Both projects are expanding the developer market and have yet to release their own incentive layers.

In terms of security, we focus on analyzing several early projects with incentive protocols

Filecoin has a large storage risk. The current economic ecology of Filecoin is not healthy, and it is difficult to cover storage costs due to low fees. The continuous storage of miners largely depends on block rewards. Filecoin has reserved 1.1 billion FIL tokens, and currently only 130 million are consumed, and there is still a relatively abundant reward balance in the short term. Moreover, the total upper limit of FIL is 2 billion, and the circulation is only 172 million, accounting for less than 10%. In the future, when the balance is exhausted, or the release of tokens causes a large inflation, and the block reward cannot compensate for the storage cost, miners may withdraw on a large scale, resulting in data loss.

Arweave’s model is healthier. Arweave replaces the continuous proof of time and space with the reward of probability random check, which cleverly solves the problem of proof of storage, and can compress the cost of proof and the probability of data loss to an extremely low level at the same time. Arweave’s one-time fee, based on the current currency price and market storage costs, can be stored for at least a hundred years. Moreover, the current circulation of AR accounts for ~76% of the total supply, and the rate of mining to be mined is less than 3%, so the risk of inflation is small. What Arweave and Filecoin have in common is that they use platform tokens to pay for storage. If the overall price of the currency falls below the storage cost, there is also the risk of miners withdrawing, but compared to Filecoin, Arweave’s risk is closer to a black swan event.

The payment model of ICP adopts the form of stable currency, which is hardly affected by the economic environment. From this point of view, the operation will be more secure and stable. However, ICP has the highest storage cost, with a monthly storage price of $0.47 per GB, while Arweave is less than $8 per GB. The price of permanently storing 1GB of data on Arweave can only be stored in ICP for more than one year.

Overall, the decentralized cloud computing market is evolving into different forms, and there is no single solution. Compared with AWS, its products that meet the needs of different levels are also constantly developed. In the future, Web3 may also form a pattern of various products to meet different needs.

Posted by:CoinYuppie,Reprinted with attribution to:https://coinyuppie.com/decentralized-cloud-computing-track-research-breakdown-of-aws-competitors/
Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.

Like (0)
Donate Buy me a coffee Buy me a coffee
Previous 2022-03-07 08:25
Next 2022-03-07 08:27

Related articles