Although the data on the chain are attached with characteristics such as non-tamperable and traceable, in the environment of decentralized storage, users hand over their files to an unknown storage terminal for storage. Words are not believable. Under such circumstances, how to ensure that untrusted nodes do not do evil? In other words, how can we be sure that the distributed storage terminal has really completed the storage task?
At this time,The Provable Data Possession (PDP) scheme of storage projects is particularly important. In terms of data possession proof, users send data to miners for storage, and miners prove that the data has been stored by themselves, and users can check repeatedly Whether miners are still storing their own data.
Of course, there are various decentralized storage solutions currently on the market. In terms of providing proof of data possession,IPFS、Sia、StorjOther decentralized storage solutions want to ensure the trustworthiness of the storage end by verifying the data possession certificate of the storage end in an untrusted environment.
Decentralized storage projects that Odaily came into contact with a few days ago Lambda Founder He Xiaoyang believes that the two existing storage solutions have different disadvantages:
The storage taking IPFS/Filecoin as an example only provides storage for static files, and the file address cannot be searched through the file content. At the same time, in verifying the proof of data possession on the storage side, it is envisaged that the public initiates verification, but it does not clearly state who the specific initiator is and how to initiate it. That is to say, it does not resolve who is the validator of the data. The problem.
Storage solutions represented by Sia and Storj want to regularly challenge the storage side through smart contracts, asking them to feedback and verify information such as file fragments and hash values. However, due to the size limitation of the chain ledger itself, only the Root Hash of the Merkle Tree is stored on the chain, which can only guarantee that the data will not be tampered with, but cannot guarantee the holding and availability of the data.
Lambda wants to build a decentralized storage platform to solve the trust problem between users and storage terminals through the verification and consensus of distributed nodes. The Lambda platform has set up two network structure systems of repository and blockchain (imagine IPFS/Filecoin). The repository system is responsible for storage, and the blockchain is responsible for the access and control of the repository system. The two networks The structures are connected by sub-chains to complete the interaction.
Different from other decentralized storage projects,What is special about Lambda is that it wants to provide proof of data possession for the storage side through the consensus generated by the verifier node, so as to ensure the integrity and recoverability of the data stored on the untrusted storage side.
The way Lambda provides is: Considering the simplest scenario, prove a storage ends possession of file F on a Permissionless Store. First, fragment the original file that the user needs to store, and generate a message set m1 , m2 , … , mn for any file F , and generate a corresponding number of tags according to the file fragment calculation. The tag consists of the original file fragment and 4 The number (that is, the security parameter λ, also called metadata) is encrypted and generated.
The validator node (validator) stores metadata for subsequent generation of challenges/riddles; each storage terminal stores original file fragments and corresponding tags, and tags can help storage parties accurately extract data from original file fragments to solve puzzles.
The verifier generates a challenge puzzle and challenges the storage side. The storage side calculates according to the original file fragments and tags in hand, generates the corresponding digital vector set and feeds it back to the verifier. Finally, the verifier calculates based on the metadata and the vector and checks whether the verification on the storage side is successful. If it is successful, the transaction will be packaged and uploaded to the chain, and the storage node will be punished if it fails.
When the single-point file ownership can be verified, file dynamics, multiple copies, erasure coding, deduplication and other features can also be supported through corresponding adjustments.
In this way, the verifier needs to save metadata, and the storage side has an extra task of storing tags, but it will only increase the storage capacity by 1%-3%. When the verifier initiates verification to the storage terminal, since the two are in different network structures, the challenge request of the verifier will be remote. When the storage end does not accept the challenge or cannot be connected due to network problems, the verifier can directly judge it as failing to store.
In the whole system, the validator role is the innovative point of Lambda. In addition, Lambda also designed the Nominator and Angler roles.
Verifier: In the Lambda network for packaging and block generation of transactions, the verifier needs to pledge part of the Token. The validator role is not a single node, but a group of validator nodes randomly selected from thousands of validator nodes to jointly verify a storage terminal. Every 1024 blocks are packaged, the validator nodes will be rotated.
Nominator: With a large amount of funds, the nominator recommends one or more verifiers to make decisions for it. There are no more functions than investing money.
Fisherman: Fisherman is not related to the process of block packaging. This role is similar to the bounty hunter in the real world. Fishermen use phishing law enforcement to report malicious validator nodes. If the report is successful, they will get a one-time reward bonus.
If the data is maliciously deleted, it is especially necessary to restore the data file at this time. In terms of data recoverability, Lambdas idea is to distribute the file to B individuals (B is smaller than A) based on the data file being divided into A, as long as the lost data is not greater than a certain preset value, Data recovery can be realized.
Related Reading:
Related Reading:
Provable Data Possession at Untrusted Stores
Scalable and Efficient Provable Data Possession
Group-oriented Proofs of Storage
English versionEnglish version、translation
English versionEnglish version
English versionEnglish version、translation
English versionEnglish version、translation
I am Mo Xinnan, the author of Odaily. I am exploring the real blockchain. If you are looking for reports on high-quality blockchain projects, please add WeChat: 16601131135. Please note your name, company, and position.
