Institution Name: Florida State University Work Package Title: ASC CY5 IC Project Title: Gateway: Computational Science Portal or Interoperable Problem Solving Environment (IPSE) for ASC -- Evaluation and Extension POC: Bernholdt David E Geoffrey C. Fox Email: bernhold@npac.syr.edu, gcf@cs.fsu.edu Phone: 315 443 3857 850-644-4587 Fax: 315 443 1973 CTA or PEI: Information and Communications Project Description: This joint project, developed by OSC and FSU, will expand upon the Gateway proof-of-concept phase from the previous funding year. The ongoing goals of this project are to provide secure, transparent access to HPC resources for DoD scientists, providing a browser-based entry point to all MSRC resources, and furthermore, facilitating the use of those resources, allowing the researcher to concentrate more on the science of their problems without requiring specialized knowledge of such things as job submission, access to mass storage, etc. In the initial phase, the feasibility of this approach was demonstrated for CCM at Supercomputing '99. In this next phase of development, the goals are to build upon the success of the previous phase by integrating Gateway into ASC working environment, getting it into the hands of working CCM researchers for thorough testing and evaluation. It is also crucial that other CTAs begin the development of their PSEs. We must also evaluate other metacomputing activities such as the Globus toolkit, Visbench, NetSolve, and JINI to determine the feasibility of their integration with Gateway at ASC. Finally, Gateway makes use of a number of rapidly changing technologies, so it is crucial that we remain abreast of current developments in web technologies so that we build a flexible system capable of changing with the times. To do this requires that the system be built using specifications (CORBA, XML) from recognized standards-producing entities. Thus it may be necessary to revisit aspects of the Gateway design. FSU's contribution to the project includes the continuing development and maintenance of WebFlow, the Java- and CORBA-based middle tier that serves as the glue between the frontend and HPC backend. Also, because the CTA specific frontends are built from WebFlow components, we are responsible for maintaining a viable interface between the middle tier and the front end. Finally, FSU will set the lead in designing more general science portals that will subsume the current PSE. Project Objectives: The primary objectives of this phase are to have Gateway thoroughly tested and evaluated by CCM users, other CTAs, and ASC security staff, and to begin integration of Gateway into the ASC working environment. It is also the goal in this phase for other CTAs to begin development of their own frontends for Gateway. Finally, with the support of ASC, integration of Gateway with the Globus toolkit will begin. It will be crucial that FSU works closely with all of the above entities to insure that the goals are met. ASC security and systems support must be included in the evaluation and transitioning process. Users from CCM and other CTA developers will undoubtedly ask for middle tier services that have not yet been developed. We also recognize the need for providing information infrastructure and generalized queuing mechanisms. These are provided by Globus, and we will work closely with ASC systems staff in getting that integrated into the ASC environment. Deliverables: FSU will provide maintenance and support of WebFlow for early CCM users. We are also responsible for all middle tier development activities. To that end we will work closely with other CTAs to develop their PSEs. This work will include the development of specialized WebFlow metatools and visualizations that can be integrated into the Gateway toolkit. Finally, we will integrate Gateway with other metacomputing tools (principally Globus, other tools will be evaluated). Customers/End Users: All users within the HPC community, both DoD and otherwise. Benefit to Warfighter: The Gateway/HPcc concept offers a promise of dramatically extending access to HPC-like resources by warfighters requiring HPC resources for their desktop environments as well as researchers while minimizing the need to purchase high-dollar mainframes. This could have major implications for the Processing Level 3 (PL3) hardware purchases being planned at ASC and elsewhere. Potential FMS applications are directly relevant to the warfighter. Other benefits include reduction of RDT&E cycle time, etc. Project Dependencies: Continued progress of the Gateway project depends on a) the quality of work done in collaboration with other CTAs, b) the demonstration to ASC security that the transparent access HPC resources does not pose a security risk, and c) the support of ASC systems personnel for third party software. Risk Element: The primary risk is that FSU, during it's transitioning process, will not be able to provide uninterrupted high-quality support for WebFlow maintenance and development. Dr. Tom Haupt, the primary developer of WebFlow, may not be involved in the project during this next phase. This is not foreseen to be a long term problem, but there may be short- term difficulties as Dr. Geoffrey Fox builds up his research group at FSU. The secondary risk is that the technologies involved will lead to dead- ends, requiring (perhaps major) reworkings of certain aspects of Gateway. We have already observed this in the proof of concept phase, as web browser technology has entered a surprising developmental lull, producing unexpected problems transitioning the SC99 demo to the kerberized environment at ASC. More generally, there is a risk because Gateway makes use of Web technologies that primarily use Public Key Infrastructure, rather than Kerberos, for security. Currently, we are dependent on a single vendor to provide a kerberized version of CORBA. The requirement for using Kerberos also can lead to unexpected roadblocks in development that can delay deliverables.