Decentralized storage, technically different from distributed storage.Decentralizing storage is a need for a more secure, more trusted, and more controllable storage in a more decentralized, less trusted network environment.A straightforward understanding can be thought of as the Bitcoin that can be stored.Here, it is important to note that the block chain itself is not used for storage, and the block chain itself is a ledger for recording transaction information.At least the current blockchain technology cannot be used directly as storage, and the future technology development can be, this depends on the demand.At present, it is not necessary.
This is the biggest problem I've faced in the past year, when I've been promoting decentralized storage.For more people, the need to de-centralize storage is questionable, as it is a pseudo-requirement."Why should my data be stored there?"Questions like this are endless.So here I'm going to talk about why we need to centralize storage.The objectives of decentralizing storage are:
Safety, reliability, performance.
First, safety.It refers to the fact that the data in the centralized storage system is absolutely safe and has very strict privacy protection.This is the biggest difference and advantage of centralizing cloud storage.In order to achieve security, each user's data is encrypted, segmented, and has multiple redundancy in the network node.These nodes may be distributed in the United States, Japan, Europe, South America, etc.Like a digital currency, only the person holding the private key can get the data, decrypt the data and view the data.In addition, hackers are not able to know which data corresponds to which users.Even if the hackers find the data, they can only hope to see the ocean.Security at this point, the current user does not seem to care, does not have the need, but does not represent the future also does not need.Sometimes the market needs to be cultivated and the product needs more patience.The recent leak of Facebook user data has had a huge impact.This is a typical scenario of data security.The user data needs to be better managed, and on the decentralized storage, the authorization mechanism can be realized through the smart contract.In the condition of user authorization, model training based on user behavior is completed.Therefore, decentralization storage is not only introduced into the block chain technology itself, but also the incentive ecosystem.
Second, reliability.The current cloud storage through the efforts of the major manufacturers, has achieved a high level, very strong.Of course, the investment is huge, and there are huge security risks.What about the reliability of decentralized storage?Technically, there is a problem of network size.Decentralized network storage, the greater the higher reliability, as the scale of the user's recognition, network is more and more big, or even can be comparable to the Internet, the reliability of data is close to 100%.And privacy security, control in the user's hand.There is a risk point, which is that when the number of early nodes is small, some measures are needed to ensure the stable operation of the network.
Finally, performance.Just think of BT, the electric donkey.If you understand the technical principle, it's almost the same.Through the technique of erasure code, calculate the appropriate redundancy to complete the tradeoff between performance and reliability.This should not be explained too much.
3. How to implement decentralized storage?
Decentralised storage implementations are divided into two parts, storage and chain.The storage part is already a mature technology, with reference to BT, the electric donkey and other technologies, taking into account the reliability and performance, as well as the consumption of resources, sufficient to meet the demand.The key point is the chain.Expansion is why you have to have chains.
It is also an old question of how to apply blockchain technology as a support.I firmly believe that decentralized storage, is a very fit, landing scene.
What does the blockchain bring to storage?Decentralization, credibility, privacy, control, and the most important incentive mechanism.If I were to design a decentralized architecture, it might look like this.
Decentralization is critical here.Think this privacy protection, think can control.So, in this scenario, we want to be more decentralized, not multi-centric -- so this determines the implementation of the consensus mechanism.
Trust, privacy.They are all based on decentralization, and are considered in cryptography.Now many people are talking about privacy protection, and even put forward the idea of "distributed privacy", how to realize -- block chain.
As we mentioned earlier, the size of the network -- that is, how much storage nodes have a great impact on the effect, how do you make people contribute to storage and network resources?How to ensure stable and reliable operation?The core here is the design incentive mechanism.There are a lot of difficulties here.At present, I have seen several storage systems: IPFS, Sia and Storj do not solve this problem very well. It is necessary to balance the interests between multiple users, at least three parties.The following items are briefly analyzed.
Therefore, the design of "Incentive Layer" above is the core.The "Smart Contract" module is also noted above.To centralize storage, do you need this module?Of course!Many core functions rely on smart contracts.For example, users should download their own files, provide storage service users to prove their ability, and so on.
To centralize the storage implementation, so much to talk about.There's a lot to think about in detail.Welcome to the planet of knowledge and explore more details.Here to give you advice, first read Filecoin, Storj, Sia white paper, and these nodes are running, too, is very fun, know how to store, mining and hard disk will be like the GPU sold out of stock?
Here is a special mention of IPFS, which is ambitious.It's not just storage, but more importantly, it's possible to replace HTTP.In addition, the overall design of Sia is better, but the design of the incentive mechanism is not ideal.The ecology is not easy to turn quickly, not easy to expand quickly.
4. From the perspective of storage development.
The previous architecture diagram is simple, but the amount of information is really large.If you look closely, there is a consensus module, the core module of the block chain.Now there are PoW, PoS and DPoS to meet different scenarios.Have you ever thought about the nature of the consensus algorithm?Like PoW, what is his nature?PoW is decentralized, relying on the large consumption of power resources to enhance the reliability of consensus.The key point is resource consumption, so can this resource be another resource?For example, can there be some new consensus ideas for storage resources, such as network resources, such as time resources?I'm just going to mention a little bit about Genaro.I have done a simple understanding before, and the consensus algorithm has some innovation, which is based on the storage +PoS mechanism, and has made some tradeoffs.The specific algorithm you can look for the white paper.I think it's a good direction, there's more room for imagination.It's like the Rank of the nebula, a nice start.
沒有留言:
張貼留言