Research customers who do not retain state data access requests, can play an important role for Ethereum, however, some fundamental limitations do not allow to apply them in practice.
At the conference Unitize co-founder, Ethereum Vitalik Buterin presented the actual picture about the latest research and talked about the problems of transition to such.
Simply put, customers without saving the data allow nodes to fully participate in the verification of transactions without the need to store more than the early history of the blockchain. In the state of the data reflects current balances, codes of smart contracts and their corresponding data.
New blocks each time, make only small changes, but they all checked around the blockchain, which ultimately ineffective process. Clients are stateless are in the focus of the initiative ethereum 1.x, the purpose of which is to maintain the existing blockchain. But studies are also important for ethereum 2.0:
“In the context of sharding in ethereum 2.0 clients and stateless are basically mandatory, since the nodes quickly change position between the different shards”.
Buterin also said that customers without saving the data studied in other blockchains including bitcoin. These customers rely on cryptographic methods only to compute the state changes and checking them, not trying to write it to memory. However, cryptography is not yet perfect.
The problem of creating evidence
One of the modern solutions relies on the Merkle proof for the inspection of the data through the concept of the witness, containing the modified information. However, this approach has several disadvantages associated with the current inefficiency of the network and the growth of each block to 405 MB.
The size of the witness may be reduced to 2MB and even to an average of 600 KB at the expense of some optimization. However, it would be above the current size of the block in Ethereum is 50 KB.
Acne Buterin see the alternative to Polynomial Commitments, verification system, relying on a polynomial function to represent the data. Due to some cryptographic properties, they can use only one “small witness, to prove a whole bunch of data,” says Buterin.
However Buterin also said that this approach has a serious problem. Evidence Merkle easily upgraded due to their tree structure, but a polynomial function, require a complete change in the curve, and, accordingly, the calculations would be costly.
There is also a solution combining polynomial commitments in a tree structure.
Research is needed
Buterin said that every potential solution has its own problems that require additional research. Concluding his speech, he said:
“There are many interesting arithmetic methods that allow to reduce the size of witnesses so that the additional data you want to load customers, will be actually not very large. But research and improvements are still needed; and we welcome assistance from the academic community.”
It Buterin suggests that customers store data in the near future in Ethereum does not appear, since everything rests in cryptographic research. Transactional capabilities of Ethereum while also deliberately restricted for fear to make the blockchain is too heavy for conventional devices.