• No results found

English. Blockchain-based Distributed Supercomputing Platform. Ali White Paper

N/A
N/A
Protected

Academic year: 2021

Share "English. Blockchain-based Distributed Supercomputing Platform. Ali White Paper"

Copied!
24
0
0

Loading.... (view fulltext now)

Full text

(1)

Blockchain-based Distributed Supercomputing Platform

Ali White Paper

(2)

Ali navigation sequence

1. background 2. introduce 3. Technology

3.1 Basic prospect objective 3.2 Architecture

3.3 Blockchaintechnology uses 3.4 Application examples

4. Blockchain Token

5. The development route 6. Safety level security 7. Service plan

8. Disclaimer

9. References

(3)

1. Background

Today, individuals or computing require the resources of high performance computing. The main ones are Amazon , Google, Microsoft, and centralized cloud service providers (with proprietary accounting policies and pre-planned resources such as IBM's closed network tasks when the company or separately purchase and use high-end high- performance workstation computers. In particular, there is a growing demand for a learning machine to run a large number of application models and high performance computing power to process large amounts of data in application development or in science and industry . However , in recent years, technology continues to provide demand due to the outstanding performance and rapid development of personal computers.

There are successes in personal computers , big data analytics and machine learning . There are successes in personal computers, big data analytics and machine learning. However, the tasks that use these high-performance computers are in the state of browsing , simple word processing and computer games , and most PCs have idle resources . However, the tasks that use these high-performance computers are in the state of browsing, simple word processing and computer games , and most PCs have idle resources . Therefore , we believe that the recognition and utilization of more PC resources need to be provided at low cost , is a new form of high performance distributed computing platform.

On the other hand, we see that new technologies provide a new way to implement distributed Block chains. Such as Etherium . In particular , Etherium enables developers to create smart contracts that run in a block chain structure. By using this approach to design services such as predictive marketing, fundraising, web hosting, general contracting , financial services, distributed data processing and execution , it creates a large number of use cases.

Wearedevelopingaplatform,tobuildaSharedbasedontheadvancedtechnologyofthepersonalcomputeridle processing resources required in the project (application ) level of high performance computing resources distributed desktop computing system can at a relatively low into the finish, through interpersonal sharing of the

(4)

ALI is a platform for building horizontal distributed desktop computing systems to handle multiple projects by sharing the idle processing resources of personal computers. The platform includes resources not only for personal computers, but also for smartphones carried by individuals. A; I is a computing resource owner (" requestor ") who rents some computing power from the computer and configures it to run the application where a person or organization (" vendor ") wants to connect to a peer -to-peer network (P2P) over the Internet on a personal computerandsharesitwithanapplicationthatrequireshighperformancecomputingpower.Thisconfigurationis useful for projects that typically require a lot of work and long processing times. ALI implements a way to organize the necessary processes from the requestor's application and execute them efficiently on distributed computing resources . This approach replaces expensive cloud computing services , reduces computing time , and makes complex applications such as scientific computing and machine learning easier and cheaper to use. Ali does not have to pay for the use of existing cloud computing because it builds high computing power by collecting and sharing resources from personal desktop computers. In addition, due to distributed computing architecture, certain organizations do not have individual control to monopolize and manage resources.

Distributed Computing

Saving the money while you sleep

We plan to use our own token based on the ethernet-based block chain to provide value for distributed desktop computing to requestors , vendors , and investors who participate in the platform running platform ecosystem . Tokens are used as a payment medium to handle distributed computing services in the platform ecosystem, and requestors and vendors can use them to exchange for the use of computer resources. This means that users who

need high performance computing power can use Tokens to solve computing performance problems , and

providers connected to the Internet can directly benefit from making their computing resources available . Therefore, ALI computing resources providers and their ability to calculate each application request and provide all these need to be involved in a liberal use of a platform of sharing the distributed computing ability, the level of all participants of convenient transactions encrypted money (cryptocurrency ) to build for the future of computing power sharing ecosystem

(5)

3.Technology

3.1 Basic prospect objective

This number is gradually increasing for the application of scientific computing and artificial intelligence , deep learning and high performance computing capabilities to handle the demand for big data at such a high level (high level) I know through the media. However, a common way to address these needs is to leverage cloud computing services or to build and deploy your own high -performance computer infrastructure . In particular , the infrastructure of high-performance computers is complex and costly, requiring professionals to operate them and separate human resources for administrative operations . In addition , traditional cloud computing is a closed network , proprietary billing policies , and preplanned resource allocation that cannot be deployed in a fully distributed infrastructure . In order to solve these problems , we need to achieve a breakthrough distributed computing resource sharing architecture, which realizes calculated resource management and reduces the cost of computer infrastructure by using block chain technology . Our system will help you automatically detect and deploy all necessary sources of computing resources, including applications, data and computing processes, using Block Chain -Based Distributed Applications (DApps ) costs , using the appropriate computing resources . The distributed application approach is now widely used, reliable and flexible enough to reduce the time and cost of implementingablockchainplatform.

(6)

Distributed computing is a distributed processing model that uses the processing power of multiple computers connected to the Internet to solve huge computing problems . Distributed computing can take advantage of multiple networks of computers , each of which performs a portion of the overall operation to achieve faster results than using a single computer . We believe that to make full use of these idle PC processing resources proved, when ARPANET was also known as the pioneer of the early 1970s, the first bright phase of the Internet network , international projects of various applications , such as biology , aero meteorology , artificial intelligence , mathematics , and cryptographic progress, we are a non-profit project, For example, SETI@home, distributed.net and BOINC (Berklee Network Computing Open Architecture), Folding@home is a nonprofit funding project.

In addition, the model of distributed computing processing using the development of the personal computer refers to the grid computing technology in the desktop network, which covers the existing high-performance workstation computers as a solution for building the basic infrastructure to share computing resources for the personal desktop computer . Grid computing is a kind of distributed distributed architecture , which combines the computing power of many computers to build virtual supercomputers to solve complex problems such as high computing and high capacity processing.

In the networked grid computing model , a server or PC runs independent tasks and is connected to the Internet or a networked network. In distributed computing, different systems in the same network share one or more resources . The advantage of grid computing is that it provides transparent access to resources to increase user productivity and thus faster completion

(7)

We will investigate the possibilities of using mobile devices (such as smartphones and tablets) in networked grid computing /distributed computing and we will configure how to apply the above technologies to mobile devices.

Theperformance ofthelatestsmartphones isgrowingfastandsteadily,andthefactthattheyarealwaysonwill bethemainreasonwhysmartphonesbecomedistributedcomputingplatforms.

In general , to run a distributed application on a heterogeneous mobile platform and a variety of network topologies , it must be abstract enough to be retrieved from the underlying platform . Deployment will be about implementing platform-specific algorithmic efficiencies and creating different versions for each platform so that they can be executed using abstract virtual machines in cross-platform technologies such as Java. In order to effectively leverage the distribution of tasks across multiple mobile devices as our computing nodes, a Java-based opensource high-performance distributed computing platform hasbeendeveloped ,Ogsi.net,for example,uses the architecture and technical architecture of the University of Amsterdam in the Netherlands and refers to the mobile platforms of Boinc and Folding@home.

Task, the network grid shrinks over time to consolidate many processors into a cluster. Provide scalability and flexibility to provide computing power where it is most needed. We refer to the Desktop Grid open source platform (XtremeWeb), which brings networked Grid computing technology to Desktop computers.

The open source platform is designed to help you build your own network grid, desktop computer, university, enterprise or voluntary, Internet-based network resources on a PC. XtremeWeb Development is based on desktop networkgrid,globalcomputingandpeer-to-peernetworkdistributedsystemsforscientificcomputingand advanced application processing. Like other large distributed systems, the platform uses remote resources (PCs, workstations, and servers) connected to the Internet and Intranet network pools (resources) that allow participants to collaborate by providing idle computing resources.

(8)

ALI allows the requestor to run the application on the machine that joins the platform. Workers connected to the Internet network simply create an environment where they can automatically participate in projects after downloading and installing executable programs and registering for accounts. The program only uses the free area on your computer, so you can adjust it to have a minimal impact on actual computer utilization. Ali will support devices that provide more value to multiple platform participants and process block chain computing resources at a lower cost. This will be designed so that appropriate compensation is achieved through transparent quality management based on the contribution of different providers and requestors , to ensure that providers are transparent about the reliability of providers providing services and high quality services

(9)

3.2 Architecture

We will design a compute node that controls the components of a personal computer . Devices such as CPU , GPU , memory and storage , and configure each node to modularize its application processing . Each node runs in the virtual machine environment and is implemented according to Docker's container manipulation methods so that it can run separately from the computer 's operating system. The storage device (disk drive ) that stores the application 's data will be implemented with reference to distributed data storage solutions such as IPFS, Storj, and SIA. Computers on each node can be run by remotely downloading and storing applications and data for service operations.

ALI configures the computers on each node to run on a horizontal distributed network. This means that there is no centralized control over the collection of nodes, as implied by the concept of block chain, where each node will autonomously access the services and configure the platform ecosystem, while controlling its own process control.

ALI implements a single resource scheduling algorithm that can manage multiple resources to reliably and efficiently process applications on multiple computer nodes. This is basically a complex program model that handles data security, management, and recovery.

(10)

< The platform architecture >

• Grid computing core

A core function of distributed computing that uses personal computer resources to handle computing tasks

• The virtual machine(Docker Container

• Hybrid peer-to-peer networks 


Actasallcomputerclientsandserversonthenetwork,regardlessofthetypeofcomputer

• SmartcontractsforEthereum

A stand-alone, self-contained application that can be deployed and automatically run on each computing node without interference from individual operating systems to the virtual processing system

A system for transparent and secure transactions, contracts and payments without the need for intermediary services

(11)

< Base Architecture (Layered Grid Architecture) >

1 Infrastructure Management Fabric Layer All computational resources, processors, disk storage, operating systems and software

2 Security Management Connectivity Layer Performs authentication and authorization of users

Secure access to resources and services

3 Resource Management Resource Layer Resource monitoring, allocation and management

4 Job Management Collective Layer Access resources and Job scheduling of various jobs 5 Application Management Application Layer Actual Applications and tools

(12)

1. Fabric Layer

✓ Provides an inteface for ALI.

✓ Shared resources

✓ Provides functionality for actual resource management, as well as the ability to query resource status. Storage system, cluster.

2. Connectivity Layer

✓ Includes a communication protocol to support network transactions across multiple resource usage ranges. Contains security protocols for authenticating users and resources.

3. Resource Layer

✓ Manage individual resources. Use the functionality provided by the connection layer and directly invoke the interfaces available in the

structure layer. Because it is responsible for access control,

4. Collective Layer

✓ Handle access to multiple resources. Resource discovery, multi- resource

task assignment and scheduling, and data replication. It can be targeted

for a variety of purposes and a variety of

✓ (APP) Consistency Control, Copying Selection, Job Management, Virtual Data Directory, Virtual Data Code Directory. (Generic) copy order

5. Application Layer

✓ A group of applications that run in a virtual organization and use a network grid to calculate the environment. (DisciplineSpecific Data Grid

Communication, Service Discovery (DNS), Authentication, Authorization, Commission

It relies on authentication performed as part of the connection layer.

Data access, computer access, network performance data access.

herefore

The protocol is configured to reflect the various services that can be provided to the virtual organization.

Recording, replication management, co-allocation,certificate authority metadata directory.

Application)

(13)

3.3 block chain Utili z ation of technology

Block chain is based on cryptographic algorithm design, such as existing public and private keys on a P2P-based horizontal network and distributed ledger technology, to ensure the integrity of books. ALI is implemented using block chain technology to record all metadata related to the processing of the computing power requester's application and manage work transactions through it. This provides a consistent and reliable environment because no one contributor can manipulate their work. In addition, Ali has developed its own token based on the Ethereum network block chain to provide users and investors with the value of distributed computing , and to provide platformparticipantswithamediumforexchanging requests andservices forprocessingservices.Itwillbeused by vendors and requestors to compensate for the use of computing resources.

(14)

3.4 Application examples

We understand the benefits and requirements of distributed computing systems . Rather than moving the resources needed by users from one computer to another, it is important to be able to process these distributed computers faster and more efficiently. This makes it possible to use an environment with high computing speed, easy system scalability, high availability and reliability, and most importantly, a good cost performance advantage

1. Projectsforscientificcomputation

ALI can be used to perform scientific computing tasks that require complex and fast computational performance . In general , scientific computing is mainly used to understand mathematical models using computers and to understand the object of study and to arrive at results . For example , academic research fields such as physics , biology , and chemical informatics , or social research purposes such as drug development, social statistics, climate prediction, and cryptanalysis, all of which requirelarge-scalecomputingcapabilities.

2. Deeplearningmodeldevelopmentproject

In recent years, artificial intelligence and machine learning in deep learning are the most active and one of the fastest developing fields . We have already seen the need for a massive computing infrastructure to develop and implement deep learning models, as we saw with AlphaGo. ALI uses this distributed computing power to implement multi-process environments that can learn deep running algorithms , such as Convolutional Neural Network (CNN ) Recursive Neural Network (RNN ) and Enhanced Learning , for rapid model creation. We are also considering developing a solution (API) to enable TensorFlow , Theano, and Caffe in our distributed computing environment (currently actively using a deep-run framework).

(15)

3. Bigdataanalysisproject

Big Data is a technique for extracting value from data , including large sets of structured or unstructured data, going beyond the capabilities of existing database management tools, and analyzing and predicting results . Big data has been prominent for several years because of its potential to provide information in all fields , including politics , society , economy , culture and science and technology.

This requires a computing infrastructure that can quickly compute and process large amounts of data, such as deep learning . Most big data analysis techniques and methods are used in existing statistics and computer science in data mining , machine learning , natural language processing , pattern recognition , etc., as well as deep learning in the case of computing and processing large amounts of data on the basis of computing . It is necessary . Especially due to the increase of unstructured data such as social media, text mining, portal mining, social network network analysis, cluster analysis and other technologies have been widely applied in the analysis technology , and distributed data processing solutions such as Hadoop NoSQL technology are used to make it more flexible and fast.

4. Computerimageprocessingproject

In the field of computer graphics, computer generated images (CGI) are movies, television programs, wide advertisements, simulators, simulations, and three-dimensional computer graphics. CGI software for use on computers continues to evolve and become easier to use , enabling medium -sized businesses and unskilled individuals to produce professional -quality computer graphics content . However, this is also a limitation, namely the computer performance required to be equipped with a high performance GPU. ALI is configured to perform computer graphics processing in a distributed computing environment so that users can run projects quickly and cheaply.

5. Lifescienceresearchstartsfirst

Biological research, used to improve existing therapies and develop new treatments for disease, is one oftheareaswheredistributedcomputingisutilized.Gpugrid.netisanon-profitdistributedcomputing project that does biomedical research through atomic biomolecule simulations. It is a supercomputer- grade computing (GPU) system made up of a large number of graphics cards (GPUs), and we are doing some tough tasks that require power. These researches have become the necessary research fields of university research institutions and biological industry , and distributed computing environment will

(16)

4.Block chain token

Ali uses its own tokens based on the Ethereum block chain to deliver the value of distributed computing to users and investors. The token can be used as a medium for platform participants to request and exchange processing services. Providers and requestors use tokens to compensate for the use of applications and computer resources.

After the application catalog and transaction framework is implemented, Tokens need to handle other processes.

As mentioned in the application examples , almost any application that requires high capacity /high performance computingpower,suchasscientificcomputing,deeplearningandbigdataanalysis,canenabletheirprojectstobe handled using the Conun platform . Any person or organization connected to an Internet network can use their ConunToken to solve their own computing capacity problems , and all Internet network users who provide their computing resources can benefit directly or indirectly

Alitoken is based on tokens published as a Block chain . The design of tokens follows the standards for token implementation that have been adopted in many places, and uses Ethernet network smart contract functionality tocreate ALItoken .Applicants wishing toparticipate inALI can purchase and keep tokens during pre-sale or crowdfunding . We will continue to pre-sell the tokens and offer the tokens to participants first. Tokens offered through pre-sale services are safely converted to the ALI default token after the ICO.

(17)

5. The development route

We will add and assign newly configured functionality at each development phase, and we will test the network services first to test the project. The initial work is done independently of the ICO and may be done faster, depending on the level of future ICOs.

■ Phase 1: Platform Development

At the early stage of ALI development, platform based technology research and development was the main focus. At the same time, we develop Conun's block chain core and token, Ethereum smart contract for token storage and transaction management.

A. Base platform definition

We plan to develop a step-by-step strategy to build a horizontally distributed desktop computing platform. ALI will be a very complex project, as we will need to develop the architecture of the new configuration with reference to the open source solution currently under development.

(18)

■ Phase 2: Alpha Version Deployment

This step implements the desktop web grid system that can be the core of the ALI platform. Develop an initial platform to implement the deployed version of the user application and run the service over the test network.

A. Design and implement the core of the desktop grid system.

B. The point-to-point network architecture is the foundation of the table network grid system. Its designislikeatablemeshgridsystem.

C. File transportation & Distributed data storage management :

D. The smart contract application implemented in Phase 1 implements additional extensions added to the distributed computing processing environment.(Ethereum Smart Contract – Phase 2)

E. ALI implements authentication programs for application use and information security-related

tasks.(Authentication & Security)

F. Implementing the user application needs to define the use of the application on the platform (ALI Task API)

■ Phase 3: Market version Deployment

In this step, configure a real-time network environment where the ALI platform can actually run and implement a deployed version that can handle distributed computing requests using Tokens. On-site networks are available to everyone in the industry, including individuals and research institutions and businesses. According to the practical application to

(19)

A. The application of category management by category (gradually) may start in the fields of scientific computing and deep learning. (Application Category Management)

B. By strengthening ALI's level of authentication and information security, and by implementing a system to manage community activities, the platform's effectiveness is managed based on user engagement. (Security & Reputation Management)

C. Build a management system that monitors ALI's user activity and job processing status. (the Resource Monitoring)

D. Conunsupportswebservicesthatcancreateandruncommunities.(CommunityNetwork)

■ Phase 4: Mobile Cluster

By developing the mobile grid system architecture and framework to support distributed computing to work on mobile devices, they also form a single network integrated with the existing desktop grid platform.

A. Develop an architecture and framework that works with multiple mobile devices and uses them

as compute nodes.

B. Build the mobile application and develop the user interface.

C. Connect the desktop network grid system with the mobile network grid system.

D. Now start building environments that can leverage computing resources on individual users' desktops and mobile devices.

Project classification of the platform. This step can be subdivided into several detailed steps. This is to avoid many of the wind risks that can occur when multiple large application projects are running simultaneously.

(20)

6. Safety and Security

We have implemented anti-corruption policies and systems for risk management, as well as security policies that can detect, isolate, block and weed out dishonest nodes from all nodes participating in the platform, authorization and authentication systems that will be designed and operated to protect systems from malicious hackers while maintaining privacy.

Distributed computing environments are complex structures that allow remote access to computing resources, such as desktop computers, mobile devices, and workstations, over Internet networks. These and other environments may be at risk from malicious threats.

We have created a responsible security culture that enables all activities on the ALI platform to remain secure and transparent, thus enabling the computers of various computing resource providers to be securely protected and risk managed to protect application users' data and information from security threats.

The security system basically uses a password-based authentication and authentication system, which operates on password-based authentication and tickets to ensure the integrity and confidentiality of the data processed in the platform . We apply VO-level systems that provide integrated authorization services and resource -level system authentication systems to grant access limits to resource sets. And in the application's execution area, it will protect the user area by implementing a secure , isolated execution environment based on the sandbox virtual system.

There are many steps required to provide a secure distributed computing environment for users, and it is critical to build an environment where trusted external applications can run on desktop computers. We are well aware of the importance of our safety and security concerns and we will continue our research and development work regardless of each stage of development.

(21)

7. Service plan

We also plan to launch a related service that will allow us to leverage our Ali platform to create a second added value.

■ Cryptocurrency exchange, the continued profitability of

ALI's long and long term business and additional

trading services for Alitoken

■ Ali's password token is used to conduct business on a

Our core service area will be a horizontally distributed desktop computing system platform that will provide the ecosystem to execute all the applications that require low cost, high performance computing, as described in the application example.

(22)

8. Disclaimer

This document is intended only as a guide and provides information to assist those interested in using the Ali Platform's services

The contents of this document are not related to any investment proposal or investment proposal and shall not be construed as a contract to sell or purchase any ALI tokens. If necessary, you are strongly advised to seek advice from a legal or financial expert.

Statements,estimatesandconclusionspresentedhereinmaycontainforward-lookingstatementsthatmay produce results that are not consistent with expected results and may be technically inaccurate or

incorrect.

This document may be updated or changed, the latest version of the document will replace the contents of the previous version, and we have no obligation to notify the fact or content of any changes. It is your sole responsibility to purchase, receive and retain Alitoken and to decide whether to use or resell it through this service, and at your own expense

Accountability for profits, losses and tax assessments.

(23)
(24)

References

Related documents

I n a €urther 3 combinations very weak complementation was only detectable in the diploid by growth tests.-Enzyme activhies in wild- type homokaryons, dikaryons and

Field experiments were conducted at Ebonyi State University Research Farm during 2009 and 2010 farming seasons to evaluate the effect of intercropping maize with

Abbreviations: AE, acute exacerbation; BDI-II, Beck Depression Inventory-II; CAT, COPD assessment test; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in

The inflow boundary condition for such LES simulations is set using the turbulence information extracted from the wind farm aerodynamics simulations performed in the first step..

When tested with simulated data, the ROC curves and true positive comparison indicate that a kernel size of roughly 0.20 should show an improvement over CCA (Figure 4), but

Further experiences with the pectoralis myocutaneous flap for the immediate repair of defects from excision of head and

KEYWORDS : Data mining, Classification, Diabetes data set, Search Methods, Tree, Meta boost,