Let’s briefly talk about a few storage projects: SpaceMesh, SubSpace, AO

Author: Hu Feitong

Quietly, the spring of encryption is passing through everyone In the corner, blockchain has taken deep roots in the earth, indicating a prosperity beyond imagination. The melody of Web3 is becoming clearer and coming towards us with firm steps. In this season of recovery, the seeds of innovation burst out of the ground, the grass stretches its green leaves, and the flowers on the lakeside bloom with a smile that belongs to spring. Nourished by the spring light, the old trees added another trace of growth, and the spring bamboo shoots burst into the sky with the warm breath.

In the courtyard of AI, everything seems particularly prosperous, and technology and intelligence are intertwined to form the most beautiful scenery. The DePin grassland is full of vitality and greenery, showing how the traditional network dances with Token. This is a spring full of hope and vitality. Everywhere tells stories of innovation and growth, leading us into a more colorful new digital era.

The story of this spring is the process of blockchain transforming into Web3. DeFi based on Token transactions and derivatives is no longer the only one main character. As a new generation of Internet, Web3 will inevitably lead the construction of a new generation of infrastructure. Whether it is AI or DePin, it talks about new computing networks and storage networks, all rollups, new consensus innovations, and new computing paradigms. A higher-performance computing architecture with these foundations will make the transition from Web2 to Web3 feasible.

(It’s too far off, take it back?) The infrastructure of the Internet is nothing more than computing, networking, storage, and the protocols and applications built for it. Today, while drinking tea, we talked about a few new storage-related projects.

In fact, PoC projects are not simple. A broad definition of PoC should include all projects that reach consensus through Proof of Capacity. As I mentioned in the article [Blockchain Random Thoughts] Three Forms of Mining: Mining/Farming/Staking, currently the only thing that can realize PoC verification in a decentralized manner is storage space. , so all the PoC projects we have seen so far are storage-related projects. It can also be said that basically all storage-related projects have PoC characteristics.

SpaceMesh uses the PoSt consensus mechanism, which can verify storage space commitments and achieve relative fairness and security. However, SpaceMesh's storage space verification is not so strict. To put it simply, it only needs to be verified once every two weeks. Of course, due to P disk technology and speed issues, it currently seems that saving storage capacity is much cheaper than re-P disk, so miners (Farmer) prefers to always reserve disk space capacity for the network.

Another very important feature of SpaceMesh is that its Plot data is specified by the network protocol, so it is not user data, and it can be said that it is not (except for consensus ) useful data. This is a flaw in SpaceMesh’s narrative.

Roughly speaking, the characteristics and development space of SpaceMesh can be considered as follows:

1. Useless PoC network, therefore, cannot be regarded as a Web3 storage network. There are many similar projects, such as Burst, CHIA, etc., which do not have an advantage in the current narrative environment. When Bitcoin is strong, it is difficult to become mainstream without other narratives;

2. The consensus is simple, it is easy to participate, and it is easier to form a mining pool, so it is easier to have a mass base. As a mining project, it has its own advantages. When the market is bullish, Consensus is easy to form and grow;

3. Because it is stored and verified every 2 weeks, SpaceMesh's overall computing needs are relatively small, and It is relatively concentrated. For individuals and enterprises with flexible computing resources, the participation cost can be relatively low; but for ordinary users, computing resources may cause a lot of waste;

4. < p style="text-align: left;">The imaginative space of SpaceMesh must rely on SVM (SpaceMesh Virtual Machine) to support smart contracts. SVM is currently in the development stage and claims to support WSAM. In the fiercely competitive virtual machine market of Web3, even if SVM is successfully developed, it will still be challenging to gain market attention.

Recommendation: Continue to pay attention to development and participate cautiously according to market conditions.

SubSpace is of course also a PoC project, because it also relies on proof of storage capacity to form a consensus, but unlike SpaceMesh, SubSpace stores useful data. The impact of this difference on the narrative is significant.

At the same time, SubSpace’s consensus is similar to the Nakamoto consensus and similar to the decentralized storage leader Filecoin. In strict accordance with the ratio of storage capacity, Farmer can use storage proof to Obtain the right to produce blocks. The biggest difference between the Nakamoto consensus and the BFT consensus is that the Nakamoto consensus has more powerful decentralization capabilities and can basically expand horizontally, while the BFT consensus often does not have many verification nodes, and reaching 3 digits is already very good. . Therefore, the consensus of Filecoin and SubSpace are both useful consensus (using useful storage to form consensus), and they are also more decentralized than most current PBFT-like networks.

Different from Filecoin, SubSpace is not intended to provide a storage market open to users, but to store chain historical data. By storing chain historical data in Multiple nodes ensure security, which is different from Bitcoin/Ethereum, etc. Bitcoin's full node needs to store all historical data, not part of it. This requirement is a bit high, which is also one of the reasons for the upgrade of Bitcoin's SegWit. The current actual situation is that most Bitcoin nodes do not need to store all data. For example, SebWit witness data is stored in only a few nodes. Due to the adoption of Ethereum It is an account status model, and there is no need to store a large amount of historical data. It only needs to store status and recent data. Therefore, the preservation and verification of historical data is a problem. This is one of the reasons why DA (Data Availability) has become so popular recently. SubSpace basically does not have DA problems because it requires the entire network to store all historical data and store many backups to ensure security. In other words, SubSpace naturally solves the DA problem through its consensus mechanism.  

In addition, SubSpace adopts a model that separates consensus and transaction execution, so it is a layered architecture. This level was implemented in the initial design. Abstraction uses Domain to execute transactions and implement subnets to achieve parallel computing. This is similar to ICP, or Filecoin’s IPC.
Currently, SubSpace’s consensus layer is relatively complete, but its contract execution, Domain interaction, etc. still need to be improved. SubSpace is currently in the later stages of the incentivized testnet and is expected to launch the mainnet in the middle of this year.
Recommendation: Pay close attention and actively participate if possible.

The Arweave team recently announced the launch of the AO testnet. AO is the computing layer on top of Arweave. Adopt Actor Oriented architecture.

The Actor model or Actor Oriented architecture is a mathematical model of parallel computing, first proposed by Carl Hewitt in 1973. This model aims to solve the complexity of concurrent computing by introducing a basic computing unit called "actors". Each actor is an independent entity that can perform the following operations:

• Create updates Multiple actors: An actor can create more actors at runtime, thereby dynamically expanding computing resources and processing capabilities.

• Send messages: actors interact by sending and receiving messages. Each message is immutable, which helps To avoid many common problems in concurrent environments, such as race conditions and data conflicts.

• Processing received messages: Each actor can decide how to respond to each message received, including changing its own internal state , send more messages to other actors or create new actors.

Actor model has the following key features

• Concurrency: The Actor model inherently supports concurrent execution because each actor can process messages independently. This makes the actor model very suitable for building distributed systems and parallel applications.

• Decentralization: There is no central control point to manage the interaction between actors, which allows the system to Easily scaled and managed.

• Fault tolerance: The Actor model enhances fault tolerance through isolation and message passing mechanisms. Each actor The failure will not directly affect other actors, and mechanisms can be designed to monitor and recover failed actors.

• Asynchronous communication: Message passing between actors is asynchronous, which means that actors do not need By blocking and waiting for a response after sending a message, they can continue processing other messages, thus improving efficiency.

Looks great. However, why is a model proposed in 1973 only now implemented and used by Arweave in the blockchain world? Why have so many years of efforts to improve TPS and concurrency not considered using the AO architecture? There are many reasons for this. Speaking of which, there are huge challenges in the combination of AO and blockchain:

1. Limitations of the consensus mechanism: The core of blockchain technology lies in its consensus mechanism, which requires all participants in the network to agree on a certain version of transaction history. This need for global consistency conflicts with the locality principle of the Actor model, which favors message passing and processing among distributed actors.

2. Transaction order and finality: The order and finality of transactions are crucial because They directly affect the status and execution of smart contracts. Although the asynchronous and independent features of AO help improve concurrent processing capabilities, it faces challenges in ensuring transaction order and finality.

3. Resource and network limitations: Although AO can improve concurrent processing capabilities, in the distributed ledger In some cases, the overhead of data synchronization and consistency maintenance may offset the benefits of concurrency.

Currently we have not seen how to achieve strong consistency of the entire network under the AO architecture, that is to say It is a blockchain; I have not seen the Actor verification method. If an Actor is executed, and the entire network ensures that its execution is correct, how to roll back if it is incorrect, etc. One possible implementation is that the AO architecture based on Arweave does not need to reach a consensus at the AO computing layer. They use Arweave as a DA to record history and messages. There is no need to reach a consistent trust relationship between AOs. Therefore, it may be a weak trust network, used for some applications that do not require strong trust, such as IM, rather than currency storage and transactions, etc.

Suggestion: Continue to pay attention to technical challenges.