"Do you really know what blockchain-based 'cloud computing' is?"
Original source: White Plan
In the recent years of the development of cryptocurrency network, there are many projects featuring capacity expansion, but not many can serve as "iconic solutions", such as the isomorpous sharding of Ethereum 2.0, the heterogeneous sharding built by Boca, the side chain of Plasma, layer2 of ZKSync, Optimistic, Starkware, etc., and the cross-chain structure of Cosmos (scaling with cross-chain).
These projects are constantly looking for the most suitable expansion method based on the Ethereum and Bitcoin blockchain structure. Ethereum 2.0 has attracted the most attention, focusing on PoW switching to POS, then compressing transaction data (rollup) and shaping sharding structure (non-data sharding). This approach is extremely long and belongs to the ultimate idea of cryptocurrency network, but it cannot be denying that it is a very necessary and long-term road.
In essence, the advantages of blockchain structure are obvious, and the ceiling is also extremely obvious. If you want to innovate, if you don't break the ceiling brought by the blockchain structure, it seems difficult for the industry to move to the next step. The author believes that the innovation in the industry needs to learn from the ideas of many mature industries and mature technology systems.
One example of breaking the performance ceiling is the design of cloud computing platforms.
Blockchain's bottleneck comes from what it is best at: consensus.
The operation process of consensus is a process in which multiple parties (node devices) calculate the same data (block). For example, Bitcoin is broadcast to all nodes after a node packs a block.
Even if Ethereum 2.0 is switched from PoW to PoS, it will only speed up the entire consensus process, reducing the time for single consensus completion to increase the number of processes per unit of time. However, in the face of massive computing demand, POS is still too obvious as the ceiling.

An example diagram of the single-node limit of the blockchain
Thus, in each blockchain structure, the model shown in the figure above will appear. All the computing tasks are preempting the computing resources of a computing node, while several tasks are preempting a narrow channel.
If some application scenarios require less concurrency, the validation process can be smoothing out by increasing the computing power of a single node, changing to a faster consensus algorithm, and allocating "pass" time to tasks that are competing for resources.
Unfortunately, for many high-concurrency scenarios (blockchains do not stop at financial and single scenarios), it is bound to be blocked, slow, or even not possible at all, or it may cause other issues (such as security) due to the blockage.
To solve this problem, it is necessary to realize enough parallel processing in the task processing so as to increase the upper limit of the task processing in the network unit time.
What about the idea of scaling and parallelism in cloud computing?
The idea provided by cloud computing, the most basic requirement, is to access the network resources of the system, not multiple computing devices access, external output of resources is only the upper limit of one computing device, but after the access of N computing devices, the processing capacity of the network is increased by N times.
This is exactly what cryptocurrency networks need. Each cryptocurrency network has access to several computing devices, and the ultimate performance is limited to the structural part of the consensus layer.
Specifically, the traditional cloud computing platform has horizontal capacity expansion and vertical capacity expansion. Horizontal capacity expansion refers to parallelism, and tasks are distributed and processed by partitions. Vertical scaling increases the processing capacity of a single device, much like one way to solve the scaling problem is to increase the number of blocks.

Examples of parallel in cloud computing, data generation suitable for parallel structure, and then use the performance of GPU fast processing
However, while the blockchain structure in the cryptocurrency network cannot be changed, the idea of implementing parallelism has evolved into two kinds.
In this article, the White Project team will take six cryptocurrency projects, OASIS, Phala, Platon, Dfinity, Filecoin, and IOTA, as examples to illustrate two main approaches to implementing parallelism.
(White plan note: the order is distinguished according to parallelism depending on secure hardware and parallelism depending on improved algorithm)
When these cryptocurrency networks have the ability to cloud parallel, they will assume the expectation of solving many problems left over by the Internet in the future.
The above projects can be divided into two approaches to solve capacity expansion and parallel problems.
The first is represented by OASIS, Phala and Platon. By connecting the trusted computing hardware to the network as a computing device, the hardware device has high computing power and security ability, which can give security to the computing process and stored procedure. And these individual devices (or clusters) can independently undertake independent processing work, so that parallel and secure computing can be realized outside the consensus layer, which can be summarized as independent trusted computing.
Secondly, represented by Dfinity, IOTA and Filecoin, new algorithms are developed at the consensus layer to change the process of block transaction confirmation and realize parallel verification, so as to improve the task processing capacity on the chain. Then, the computing power and space of a single computing device are increased through the shaping of scalability, that is, the vertical expansion of cloud computing.
The specific breakdown is as follows:
1. Build a good consensus layer first.
First, cryptocurrencies need a general ledger, which exists in the consensus layer. OASIS, Phala, and Platon all split the consensus layer from the computing layer, with a separate consensus layer on top of the computing device, namely a blockchain network that runs high-speed consensus algorithms through the computing device (or cloud).
However, it is worth noting that OASIS and Platon have relatively obvious hierarchical concepts, while Phala's hierarchical concepts are not obvious. Its design detail is that it has independent rules for off-chain computing devices.
In order to ensure the stability of the consensus layer, this layer of OASIS chooses to build nodes through organizations and enterprises with high trust in the industry, and the nodes communicate with Tendermint algorithm to quickly form general ledger books.
Platon's nodes are also built by partners, and it uses CBFT algorithm like BFT algorithm to optimize the efficiency of ordinary BFT algorithm.
Phala connects the computing node with TEE (called Gatekeeper) to the network. Gatekeeper's TEE computing area can maintain the general ledger book, and its consensus is the same as Poka's NPOS consensus, which can quickly produce blocks.

Phala's Gatekeeper (in the middle) maintains the ledger book
Outside the consensus layer, they bring computing and storage into the chain or layer2. This is where parallel computing is implemented.
2. Let the computing layer implement parallel computing.
OASIS, whose computing layer is called Paratime, can be seen as a separate chain or as a runtime cluster. However, in the early days of the OASIS network, Paratime was mostly deployed in the cloud and did not fully replace the devices with TEE as the network infrastructure. As the process progresses, all Paratime nodes will be TEE capable to ensure their security.

OASIS Computing Layer (right)
Pruntime is deployed in each TEE. Pruntime communicates independently with the "consensus layer" (in theory), so the transactions processed in Pruntime do not conflict with each other. This is the parallelism that can be achieved because the Pruntime in each TEE is like a "shred". The more such access nodes, the stronger the network performance.
PlatON calculation is done in was noted for the calculation of layer2 layer, PlatON layer2 with a large number of computing devices, including custom trusted computing devices, such as complete multi-party computation programmable circuit, moreover, will complete privacy through technologies such as cryptography and zero knowledge proof calculation, etc., PlatON, also of privacy, but the use of technology is a multi-party computation or zero knowledge proof, homomorphic encryption, etc.

Modules and layering of the Platon network
To design the computing layer as a network of trusted computing hardware is to make use of the parallelism of the computing layer to expand capacity and achieve scalability. We might think that moving computing outside the consensus layer does not really mean parallel computing.
But the hardware with trusted computing and the consensus layer are "integrated" with the consensus layer because of their close connection to security. Theoretically, for the security of off-chain computing, there should be a general ledger concept or other means of controlling security under the chain. But with the help of trusted computing hardware, there is no need for this general ledger to provide off-chain security.
A comparison could be made with Ethereum 2.0, where the beacon chain is the general ledger book. If the shard has been deployed, each shard can handle tasks independently, but in OASIS, Phala, and Platon, the computing part of the shard is replaced by trusted hardware.
After splitting the parallelism of the computing layer, let's look at the way the splitting uses algorithms to achieve parallelism.
1. Develop new algorithms.
With Dfinity, IOTA and Filecoin as the representatives, the algorithm can be developed to enable parallel processing of tasks without changing the block confirmation process, thus speeding up the confirmation speed.
Here we should first mention that if parallelism is implemented at the algorithm level, the main implementation will be to change the calculation rules of the algorithm, which will also change the functional logic of the algorithm performance. For example, if the POW algorithm is changed, the logic of calculating random number, packing and broadcasting in the POW algorithm will be changed.
Part of the change of Dfinity algorithm is the consensus algorithm, which changes the traditional consensus nodes to participate in consensus calculation to select some nodes to complete consensus calculation by calculating random numbers, which is a step to speed up consensus verification. What is more core is that the selected consensus node confirms the transaction through the non-interactive BSL algorithm (node confirmation data signature feedback is carried out independently, not in combination), which means that it will not go through the process of repeated interaction between nodes of BFT consensus, and achieve the effect of similar "parallel" acceleration.

In the consensus confirmation process of Dfinity, the signature part on the left side has the parallel effect
IOTA has modified the algorithm thoroughly. Compared with blockchain, IOTA uses Tangle data structure to form a general ledger. The Tangle feature is that each transaction is attached to two previous transactions, so it completely eliminates the dependence of the original blockchain chainlike structure on confirmation time. This forms the infinite association confirmation structure of the transaction, which can achieve the parallel effect.

Trade confirmation model of Tangle algorithm
Filecoin's revisions to parallelism are parallel processing of the storage task, because the storage part of Filecoin does all the computation of the stored data, which is a very long process (by comparison). Therefore, parallelism and speed up are very important. At present, it uses the updated NSE algorithm.
What can be seen from splitting NSE algorithm is that, when processing data, the data will be divided into Windows (which can be understood as a unit) and layers. After the completion of processing, the next step of data storage and subsequent POST proof packaging will be carried out. After NSE is adopted, there is not too much dependency between layers in the processing part of the layer, so the parallel processing effect can be formed, which can be summarized as the adjustment of parallel acceleration.

The FilecoinNSE algorithm is dismantled. You can observe the layer section on the left
2. Configure the rest
We solved the parallelism problem algorithmically, so we need some auxiliary functions next.
IOTA Tangle does not have the time limit of ordinary block structure, so in order to reach a consensus, we need the help of a transaction validator to confirm which transactions form a consensus.
Dfinity improved the algorithm and matched it with subnets, data centers and containers. Subnets are similar to "shards", while data centers are the underlying network deployment of Dfinity network, which requires the participation of data centers, meaning that the basic processing capacity of the network is very strong. On a subnet, containers are built as independent units of operation, analogous to smart contracts on a blockchain, and the combined interactions of containers can achieve complexity.
After parallel processing of data by NSE algorithm, Filecoin carries out storage replication and packaged storage of spatio-temporal proof. These sections ensure consistency in the Filecoin total ledger. The rest of the development relies on the official team and the tools available in the ecosystem.
The above six cryptocurrency projects theoretically break the performance limits of blockchain in parallel, so what's left for the project next?
The author thinks about how to make these capabilities available to developers through the development of web tools. The most important goal of the use of cryptocurrency network is that it can develop dApps and decentralized business in the broad sense.
Even if the infrastructure is performing well, the infrastructure will be useless without developers making application time. Developers determine the number of applications produced in the chain, and the number of applications produced determines the value created and contained in the chain.
Just like the traditional Internet developers, the cloud computing platform has provided the developers with extremely high quality experience from the humble foundation development to the era of cloud development. Later entrepreneurs, unlike those days, still have to worry about expansion.
I wonder if today's cryptocurrency networks can follow the example of cloud computing platform "service-oriented architecture" and form a wave of development. After cloud parallel, cryptocurrency just broke through the well, can you continue to rise to the sky?
The original link
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia