Special Report: Network Provisioning
In this report we briefly discuss the networking requirements of DOE large-science applications in Section 2 to highlight their needs and scope. Section 3 discusses various details of the workshops including the composition of working groups. Sections 4 and 5 are devoted to the main technical topics of this workshop, namely provisioning and transport, respectively. In each section the problem space and basic issues are described briefly, followed by the recommendations of working groups in terms of topics of interest in respective areas. The development of the required network technologies warrants a science of high-performance networks described in Section 6.1. The transport and provisioning technologies must be transparently integrated into the applications, and such issues are described in Section 6.2. Furthermore, these technologies can be efficiently tested using powerful test-beds that support close interactions with the applications, and these aspects are discussed in Section 6.3. Although originally not intended, the cyber security aspects were discussed in Section 6.4 due to their increasing and often very intrusive impact on the provisioning and transport methods.
2. Advanced Networking Requirements for DOE Large-Scale Science
2.1 Ultra-Scale Science Environments The DOE large-science applications are quite varied in terms of their network requirements in part due to their disciplinary origins, which are as diverse as earth science, high energy and nuclear physics, astrophysics, fusion energy science, molecular dynamics, nanoscience, and genomics. The networking requirements for some of these areas are listed in Table 1. In this section, we describe these needs only briefly to highlight their general nature, and a detailed account of a number of DOE large-science applications and their networking requirements can be found in [1]. Table 1. Network requirements for DOE large-scale science applications. Many DOE applications rely on high-performance heavy-lift data transport that requires an optimal combination of network provisioning and transport protocols. For example, the network requirements of High Energy Physics (HEP) data transport applications are unprecedented: they must deliver hundreds of Gbps throughputs between two applications in near future and several Tbps within the next decade. In contrast, some other applications could require several concurrent channels for tasks to be cooperatively performed over wide-area networks by experts distributed at various national laboratories and universities. These tasks could range from cooperative remote visualization of massive archival data through the distribution of large amounts of simulation data, to the interactive evolution of computations through computational steering. In the case of remote visualization, the data must be rendered and presented on-line to various participant sites with different end-devices ranging from visualization caves through high- end workstations to personal desktops. Furthermore, the control of such visualization streams may have to be handed back and forth among the sites, while maintaining a smooth response of the distributed rendering engine. Details of two specific example applications with their requirements are provided in Appendix A.
Previous Next Table
of Content for report: Network Provisioning Home
