The software architecture of a virtual distributed computing environment

Abstract

The requirements of grand challenge problems and the deployment of gigabit networks makes the network computing framework an attractive and cost effective computing environment with which t o interconnect geographically distributed processing and storage resources. Our project, Virtual Distributed Computing Environment ( V D C E ) , provides a problem-solving environment f o r high-performance distributed computing over wide area networks. V D C E delivers well-defined library functions that relieve end-users of tedious task implementations and also support reusability. I n this paper we present the conceptual design of V D C E software architecture, which is defined in three modules: a) the Application Editor, a user-friendly application development environment that generates the Application Flow Graph ( A F G ) of an application; b) the A pplication Scheduler, which provides an efficient task-toresource mapping of AFG; and c ) the V D C E Runtime System, which is responsible for running and managing application execution and monitoring the VDCE resources. I n t r o d u c t i o nGrand challenge problems have computational and storage resource requirements that are beyond the capacities of a single computing environment. Addition-*This research is supported by Rome Lab contract number F30602-95-C-0104. ally, emerging network technologies such as fiber-optic transmission facilities and the Asynchronous Transfer Mode (ATM) enable data to be transferred at the rate of a gigabit per second (Gbps). A high-speed network of geographically distributed heterogeneous resources represents a cost-effective, network-based computing environment for solving large-scale problems addressed by grand and national challenges. New software development models and problem solving environments are being developed to utilize efficiently the network computing environment.The software development process of parallel and distributed applications can be broadly described in terms of three phases: a) application development and specification, b) application scheduling and resource configuration, and c) application execution and runtime. Most of the related work so far has focused only on one or two of these phases; only a very few projects have completely addressed all phases of software development.The first phase, i.e, parallel and distributed application development and specification phase, overwhelms most users because of the difficulty of expressing communication and synchronization among computations 131. Some text-based parallel programming environments support the data-parallel paradigm, which requires advanced compilation techniques and compilers. Most of the other environments require explicit insertion of communication and synchronization primitives within the programs, which makes programs difficult to understand. Over the last few years a number 40

The Software Architecture of a Problem Solving Environment for Enterprise Computing

“…More generic infrastructure for building PSEs is also under development ranging from fairly simple RPC-based tools for controlling remote execution [9] to more ambitious and sophisticated systems such as Legion [10] and Globus [11] for integrating geographically distributed computing and information resources. The architecture of Virtual Distributed computing Environment (VDCE) developed at Syracuse University [12] is very similar to the software architecture of the E-PSE described in section 3. However, VDCE has not modeling module for enterprise, which is more suitable for computing science than enterprise computing.…”

Visual programming support for graph-oriented parallel/distributed processing

“…VDCE [21] is based on the dataflow programming paradigm. The software architecture consists of three parts: Application Editor, Application Scheduler, and VDCE Runtime System.…”

“…Directed-graph program representations are used to specify multiple threads of control, to display and expose parallel structure, and to express communication and synchronization, which are separated from the specification of sequential computations in individual local programs. In the past decade, a number of visual programming environments have been developed [2,[16][17][18][19][20][21][22].…”

Adaptive Distributed Virtual Computing Environment (ADViCE)

“…Recently, there has been an increased interest in building large scale high performance distributed computing (Metacomputing) such as Globus 9], Legion 16], VDCE 2,4]. These metacomputing projects provide large scale applications with computing and storage power that was once available only in traditional supercomputers.…”