Domain Computing: The Next Generation of Computing

Abstract

Computers are indispensable in our daily lives. The first generation of computing started the era of human automation computing. These machine’s computational resources, however, were completely centralized in local machines. With the appearance of networks, the second generation of computing significantly improved data availability and portability so that computing resources could be efficiently shared among the networks. The service-oriented third generation of computing provided functionality by breaking down applications into services, on-demand computing through utility and cloud infrastructures, as well as ubiquitous accesses from wide-spread geographical networks. Services as primary computing resources are far spread from lo- cal to worldwide. These services loosely couple applications and servers, which allows services to scale up easily with higher availability. The complexity of locating, utilizing and optimizing computational resources becomes even more challenging as these resources become more available, fault-tolerant, scalable, better per- forming, and spatially distributed. The critical question becomes how do applications dynamically utilize and optimize unique/duplicate/competitive resources at runtime in the most efficient and effective way without code changes, as well as providing high available, scalable, secured and easy development services. Domain computing proposes a new way to manage computational resources and applications. Domain computing dy- namically manages resources within logic entities, domains, and without being bound to physical machines so that application functionality can be extended at runtime. Moreover, domain computing introduces domains as a replacement of a traditional computer in order to run applications and link different computational resources that are distributed over networks into domains so that a user can greatly improve and optimize the resource utilization at a global level. By negotiating with different layers, domain computing dynamically links different resources, shares resources and cooperates with domains at runtime so applications can more quickly adapt to dynamically changing environments and gain better performance. Also, domain computing presents a new way to develop applications which are resource stateless based. In this work, a prototype sys- tem was built and the performance of its various aspects has been examined, including network throughput, response time, variance, resource publishing and subscription, and secured communications.