元计算:零信任下的新型计算范式

Meta computing:a new computing paradigm under zero trust

互联网的普及对计算范式的发展产生了重大影响。随着新一代信息技术基础设施的不断完善,学术界与工业界在持续探索新的计算范式,以实现对算力的充分挖掘。海量物联网(internet of things,IoT)设备产生的庞大数据逐渐超出了以云服务器为代表的高性能后端的处理能力,边缘计算(edge computing)通过云边端协同缓解了这一问题,但仍存在算力利用率低、算/存容错性低,以及计算资源整合度低等困难与挑战。元计算(meta computing)是一种新型计算范式,旨在零信任基础上,打破算力藩篱,整合网络中所有的可用计算与存储资源,为各项任务提供高效、安全可靠、可容错的个性化服务,同时利用密码学技术保护敏感数据信息,确保用户数据的隐私性,保证任务结果的准确性与可靠性,最终实现“对任何一个人或者一项任务整个网络就是一台计算机”,即“网络即计算机(network-as-a-computer,NaaC)”,也称为“元计算机”(meta computer)。本文分析了元计算的三大功能目标,据此提出了一个包含云边端资源、设备管理与零信任计算管理模块的元计算机架构。在该架构中,设备管理模块将海量异构设备的资源抽象为可以自由操控的对象,零信任计算管理模块则根据用户任务需求直接调度计算资源,完成强容错的计算任务并输出可验证的计算结果,最后进行结算。本文在剖析元计算机架构与功能特性的基础上,分析了实现元计算机面临的技术挑战,给出了元计算“从局部过渡到整体”的发展思路,预测了元计算的未来应用场景,为未来元计算的落地与发展规划出合理的突破路线。

The popularity of the Internet has had a significant impact on the development of computing paradigms.With the continuous improvement of the new generation of information technology infrastructures,academia and industry have been constantly exploring new computing paradigms to fully exploit computing powers.The huge data generated by massive loT devices gradually exceeds the processing capability of the high-performance back-end represented by cloud servers.Edge computing alleviates this problem through cloud-edge-end coordination,but there still exist difficulties and challenges such as low computing power utilization,low computing and storage fault tolerance capability,and low integration of computing resorces.“Meta Computing"is a new computing paradigm that aims to break down the barriers of computing powers in a zero-trust environment.It can integrate all available connected computing and storage resources,provide efficient,byzantine fault-tolerant,and personalized services,while protecting the data and user privacy.Furthermere,meta computing can ensure the correctness of results,and realize“the entire network can be regarded as a giant computer for a user",that is,“"Network-as-a-Computer,Naac",which is also called“Meta Computer".The meta comput er architecture includes a number of function modules including a device management module and a zero-trust computing management module,as well as the cloud-edge-end device resources.The device management module abstracts massive heterogeneous device resources into objects that can be freely manipulated,while the zero-trust computing management module can transparently allocate computing resources to user tasks according to service requirements,complete the tasks with strong fault tolerance and verifiable outputs,and finally settle accounts.Based on the analysis of the architecture and the functional characteristics of meta computing,we put forward suggestions and advice on the development path of meta computing,that is,transitioning from local to global,and analyze the most-influential future application scenarios of meta computing and provide a reasonable plan for the implementation and development of meta computing in future.