TeraGrid User Portal v1.0: Architecture, Design, and Technologies Maytal Dahan * , Eric Roberts * , and Jay Boisseau Texas Advanced Computing Center The University of Texas at Austin Austin, Texas 78758 Email: {maytal,ericrobe,boisseau}@tacc.utexas.edu * Equal Authorship Abstract—The TeraGrid [1] is a grid computing project for building a world-class comprehensive distributed infrastructure for scientific discovery. TeraGrid comprises many heterogeneous systems that enable high performance computing, data manage- ment/storage, and scientific visualization, and access to scientific data collections from facilities across the country. TeraGrid uses advanced networking systems, software technologies, and oper- ations and support activities to tie these fundamental resources together into a production cyberinfrastructure for science and research. Allocations, accounting, security, resource monitoring, consulting, and documentation are among the services that each TeraGrid Resource Provider (RP) implements to meet their site- specific needs while operating within the TeraGrid environment. The purpose of the TeraGrid User Portal [2] is to serve as a launch pad for new users and a control panel for current users by integrating RP resources, services, and information into a single web interface serving a national community of computational researchers. The first version of the TeraGrid User Portal [3] addresses the most fundamental issue for integrating TeraGrid resources: extension and integration of existing, centralized TeraGrid ac- counting and security services with a clear and comprehensive production portal interface. In addition, the portal provides simple access to existing TeraGrid user-centric information and services such as documentation, consulting, allocation request and renewal, and resource monitoring. This paper discusses the moti- vation, architecture, technology, and challenges used to build an enterprise portal for the TeraGrid. It also illustrates techniques, such as load balancing, redundancy, and user auditing, that are required to ensure performance, reliability, and traceability of the portal environment. I. I NTRODUCTION The term user portal [4] is used in the computational re- search community to describe a web-based interface that pro- vides access to targeted information, resources, and services for a community of users, scientists, and engineers. A grid user portal serves as a single interface for providing this access to a set of distributed resources integrated using grid-computing technologies. Thus, the TeraGrid User Portal is designed to serve as the main interface for providing information and services to users of TeraGrid resources. The TeraGrid User Portal v1.0 presents detailed information about the specific resources on the TeraGrid, such as machine load, job queues, user documentation, and more. Equally importantly, the portal offers an easy, consistent, and unified account interface that enables users to manage the complexity of TeraGrid accounts, allocations, and distinguished names. In section three we describe the architecture and design behind the TeraGrid User Portal. Sections four and five describe the services currently being offered in the first production version of the TeraGrid User Portal and the technologies used to achieve this functionality. Section six describes the security infrastructure and section seven discusses the chal- lenges encountered while developing and deploying the user portal. Finally, in section eight the paper discusses the future capabilities that will be included in upcoming releases of the User Portal. II. ARCHITECTURE The basis for the User Portal architecture revolves around the web server. We use the Apache Tomcat HTTP web server and within the web server are the individual portlet web applications [5], authentication modules, the GridSphere portal framework [6], and the JSR168 [7] portlet container, which are all discussed in more detail in section five. The web server receives requests from clients and manages resources (such as database connection pools) used to communicate with backend services. The individual portlets each depend on backend services in- cluding the TeraGrid Central Database (TGCDB), the GridPort Information Repository (GPIR) [8] web service, the MyProxy credential management [9] server, and the TeraGrid website. Figure 1 shows a detailed architecture diagram of the TeraGrid User Portal. The portlet web applications in the User Portal make use of a common and powerful design pattern known as Model-View- Controller [17] (MVC). The MVC paradigm breaks things up into three separate pieces: the model, the view, and the controller. The architecture is designed to separate the appli- cation (model) from the way it is represented (view) and the portlet business logic (controller). This design enables rapid development and deployment of new portlet web applications. Using MVC we have developed a wide variety of interfaces to services described in the next section. III. SERVICES SUPPORTED The TeraGrid User Portal presents many different types of information and services through a single interface by