Middleware to Support Group to Group Collaboration - Research

The Access Grid technology (www.accessgrid.org) was developed by the Futures Laboratory in the Mathematics and Computer Science Division at Argonne National Laboratory with support from the DOE 2000 program, and is now in use at over 40 institutions include major DOE National Laboratories and NSF PACI Universities. We propose in this project to address some of the key middleware research issues facing large-scale deployment of high-end collaborative environments such as the Access Grid. These high-level research goals and corresponding tasks have been formulated as the result of nearly 8 years of research into advanced collaboration infrastructures. We have identified the following areas as high priority for future development: Virtual Venues ? Our goal is the development of a scalable virtual venue server that will enable the deployment of peer-to-peer based virtual venue servers that can be used to provide public and private access to millions of virtual locations, while preserving the spatial metaphors for navigation, access and resource organization. Security ? Our goal is to improve the security model of the Access Grid to support secure sessions, private venues and secure transmissions and secure applications. This security model will be developed in close collaboration with emerging Grid security and will be fully integrated with the Grid services model in the future. Workspace Docking ? Our work in this area is aimed at creating a general purpose seamless mechanism for sharing applications via the Access Grid, in particular sharing applications from personal workspaces that might be on a desktop or laptop system with groups of co-workers in an organized and transparent fashion. We call this concept “workspace docking” with reference to the analogy of docking a laptop into a network to gain access to local services. Node Management Interface ? Our work here is aimed at improving the user interface and node management environment of the Access Grid to reduce the overhead in running a session and incorporating new types of interaction devices into the Access Grid environment. Tiled Display Interfaces ? This work is focused on developing the technology for more closely integrating Access Grid capabilities with those of scalable displays and remote visualization environments. This activity leverages the ongoing work at Argonne in developing high-resolution tiled displays and other advanced display environments. It is also essential for improving the ability for the Access Grid to support high-performance remote visualization. Asynchronous Collaboration ? This work is aimed at leveraging our previous work in developing multimedia recording and playback technology by extending it to capture, synchronize, record and playback a variety of multimedia applications. Each of these areas has strong applications drivers to enable evaluation and progress in them will greatly improve the utility of the Access Grid, which even in its relatively immature state quite usable. We discuss each of these research tasks in more detail below. As mentioned earlier our goal is to release to the community software that addresses each of these important areas during the course of this project. Deliverables are detailed in the work schedule and deliverables section.

Major Goals and Technical Challenges:
Scalable Virtual Venues Service: The current implementation of the Access Grid uses a spatial metaphor to control the scope of interaction and to provide persistence. This solution offers a set of “rooms” which are virtual spaces mapped to multicast addresses. This version has not been designed to be scalable or to provide persistence services beyond simple presence. We envision the Access Grid peer-to-peer venue services operating much the same way that the web is peer-to-peer: anyone can host a server (a virtual space) and anyone on the network can visit. The goal is to create a venues service that scales to thousands of nodes, with no centralized services; one where anyone can trivially create new spaces and link them into the peer-to-peer infrastructure.
Access Grid Security: In order to gain widespread deployment of the Access Grid, the implementation must address the security concerns of its users. At the least, the system should provide as much privacy that one expects from the telephone system: for most uses, it is private enough, but one can expect that with a certain amount of effort eavesdropping is possible. The security model for the Access Grid will be based on a rigorous analysis of the security required by the Access Grid community and the expected threats to this security. In particular, in order to ensure security that can be trusted, each and every mechanism and system implemented as part of the Access Grid architecture must be evaluated in terms of its vulnerability to attack and its relationship to Access Grid security mechanisms.
Application Sharing and Dynamic Workspace Docking: AG users often desire to “share” some portion of their personal workspace (current desktop applications and data) with other AG users, nodes, or sites, both local and remote. This desire is often extemporaneous and any proposed solution must recognize and enable the ad-hoc nature of this desire. The docking step in this case amounts to migrating or launching one or more specific application clients (linked with multicast as needed) on to the Access Grid displays and attaching them to the user’s server.
Node Management and User Interfaces: We also are developing a software layer that will improve node operations through simplified user interfaces, automated node configuration, and node management functions. We will produce the Access Grid Node Architecture Specification. We will create a configuration specification for the Access Grid Nodes Hardware, Software, and Services.
Remote Visualization and Tiled Display Interfaces: We are developing the Xplit architecture to address this requirement. The principal goal of the Xplit architecture is to enable high performance access to clustered display resources (the display farm) while providing an apparent single point contact for display services
Asynchronous Collaboration Capabilities: We propose to extend the Voyager system to include the streaming data types required to capture the interactions and events that occur in the persistent spaces of the Access Grid. These include streams of control information used for distributed slide shows or Web browsing; high-resolution lossless encodings of experimental data or simulation output; streams of navigation information from distributed exploration of large data sets; and so on.

The Middleware to Support Group to Group Collaboration is support by:
The Scientific Discovery through Advanced Computing (SciDAC) Project, United States Department of Energy (USDOE).