The TeraGrid, sponsored by the National Science Foundation's Office of Cyberinfrastructure, is a partnership of people and a comprehensive collection of resources and services that enables and accelerates discovery in U.S. science and engineering research.
In addition to the computational, networking, and storage resources NCSA contributes to the TeraGrid, the center also provides assistance to scientific users, key technologies, and crucial services.
Resources
NCSA offers TeraGrid users access to high-performance computing and data systems.
Science Support
NCSA works closely with TeraGrid users, assisting them with tuning and optimizing applications such as NAMD and WRF and with architecture-specific challenges.
NCSA staff also participate in the Advanced Support for TeraGrid Applications (ASTA) Program, which ensures that scientific users are able to optimally use TeraGrid resources to achieve breakthrough scientific and technical innovations.
Technologies
- The Account Management Information Exchange System (AMIE) was developed in partnership with Boston University. AMIE provides automated management of grid resource allocations, user information, user login accounts and the reporting of usage across a distributed system spanning multiple administrative domains. AMIE is packaged and distributed via the NMI Software Repository, with Version 1.0 debuting in summer 2004. For the TeraGrid Account Management system, the current version of AMIE is used as the information exchange mechanism. There is a central "site" through which all data flows to and from the local TeraGrid sites. However, this central site does not correspond to one of the actual resource providers—it is instead a central information repository and clearinghouse for TeraGrid account management data, implemented as a relational database. Each of the TeraGrid local sites runs a version of AMIE alongside their local account management system, through which they correspond with the central database, where another "master" instance of AMIE runs.
- CluMon is a tool for facilitating cluster monitoring and management as well as providing system status information in an intuitive, graphical view. CluMon provides system administrators with a graphical view of conventional system information such as processor state, memory usage, system load, etc. End users benefit from an intuitive, graphical view of queuing system information such as queues, job state, job status, etc. CluMon is deployed and used on NCSA's high-performance computing systems as well as some systems at other TeraGrid sites.
- HDF software I/O libraries for writing and reading scientific data sets were developed at NCSA and are maintained and supported on the TeraGrid by staff at NCSA. These libraries—including pHDF5 which takes advantage of parallel filesystem capabilities for improved I/O throughput—are used by many TeraGrid scientists.
- MyProxy allows users to manage grid credentials using standard authentication methods, such as standard passwords, one-time passwords, and Kerberos tickets. It has been adopted as the authentication service for both the TeraGrid Ticketing System and TeraGrid User Portal, and is also part of the Common TeraGrid Software Stack (CTSS), providing a uniform interface to credential management services across the TeraGrid sites.
- The Partnerships Online Proposal System (POPS) enables online proposal submission for TeraGrid compute resources, and subsequent online review and administration of these proposals. The system consists of three Web interfaces, one each for proposal submission, proposal review, and proposal administration. The backend of these Web interfaces is a Sybase Relational Database running at NCSA.
- NCSA's in-house ticketing system provides a simple, standard, easy-to-use method for the TeraGrid's diverse and geographically dispersed users to report problems and a standard, TeraGrid-wide method of tracking and resolving those problems. Users simply send email to help@teragrid.org and a ticket is created from their email and routed to the appropriate person or group. The ticket system interface is browser-based; there is no software to install and the interface is accessible from anywhere. Incoming email, ticket assignment, responses to users and replies are all handled through the browser. The ticket contains the complete record of responses, replies, and notations generated by the resolution of the problem.
- UberFTP is a popular interactive GridFTP-enabled FTP client first distributed in 2003. UberFTP is part of the common TeraGrid software stack and is supported and enhanced by NCSA staff in direct response to TeraGrid needs.
- Virtual Machine Interface (VMI) is a middleware communication layer that addresses the issues of availability, usability, and management in the context of large-scale storage area networks interconnected over wide-area grids.
Services
NCSA staff, in collaboration with staff at the San Diego Supercomputer Center, staff the TeraGrid Operations Center 24 hours a day, seven days a week. Operations Center staff provide a single source of assistance and problem resolution for any problems experienced by TeraGrid users. The TeraGrid HelpDesk can be reached by emailing help@teragrid.org or calling 1-866-336-2357 at any time.
NCSA also is the hub for TeraGrid Allocations. Allocations requests of any type are initiated and subsequently completed by NCSA staff. These activities include creating new user accounts, adding users to existing projects, creating new projects, and sending login packets to users.
NCSA staff execute weekly tests to measure the performance of cross-site data transfers. Results of these tests are shared with TeraGrid staff at all sites and are used to set expectations regarding achievable transfer rates, as well as to identify problems and bottlenecks.
Staff at NCSA also manage the many e-mail lists and aliases that enable TeraGrid participants across the country to communicate and collaborate effectively.
