Aleksandar M. Bakic

Software tools for complex distributed systems: toward integrated tool environments

The demands on software tools for the design and testing of complex distributed systems are considerable. An environment that integrates domain-specific and commercial off-the-shelf tools and that supports rapid prototyping of application-specific tools can greatly increase the functionality and usability of such tools. We look at distributed computing systems as complex systems, focusing on two contemporary examples, and present an overview of online monitoring and dynamic analysis tools that support the design and test of such systems. To provide a specific example of integration and rapid prototyping, we describe an integrated tool environment that we have targeted at the types of complex systems addressed in this article.

BRISK: a portable and flexible distributed instrumentation system

Researchers and practitioners in the area of parallel and distributed computing have been lacking a portable, flexible and robust distributed instrumentation system. We present the Baseline Reduced Instrumentation System Kernel (BRISK) that we have developed as a part of a real-time system instrumentation and performance visualization project. The design is based on a simple distributed instrumentation system model for flexibility and extensibility. The basic implementation poses minimalistic system requirements and achieves high performance. We show evaluations of BRISK using two distinct configurations: one emphasizes isolated simple performance metrics; and the other BRISKs operation on distributed applications, its built-in clock synchronization and dynamic on-line sorting algorithms.

An on-line performance visualization technology

We present a new software technology for on-line performance analysis and visualization of complex parallel and distributed systems. Often heterogeneous, these systems need capabilities for flexible integration and configuration of performance analysis and visualization. Our technology is based on an object-oriented framework for rapid prototyping and development of distributable visual objects. The visual objects consist of two levels, a platform/device-specific low level, and an analysis- and visualization-specific high level. We have developed a very high-level, markup language, called VOML, and a compiler for component-based development of high-level visual objects. The VOML is based on a software architecture for on-line event processing and performance visualization called EPIRA. The technology lends itself to constructing high-level visual objects from globally distributed component definitions. We present details of the technology and tools used, and show how an example visual object can be rapidly prototyped from several reusable components.

Teaching undergraduate computer science and engineering students techniques for the design and analysis of real-time applications

Michigan State University has begun a project to incorporate embedded systems education across several courses. This paper provides an overview of the project and details specific efforts to enhance one of the courses, operating systems, in order to include special instruction of real-time computing. The instruction material developed is modularized such that its components detailing scheduling issues and laboratory project ideas and infrastructure will be available for other courses, such as the software engineering and the computer engineering project courses.

Modelling, evaluation, and adaptive control of an instrumentation system

We present results from modeling and evaluating the JEWEL instrumentation system (IS), which is being used for runtime data collection from a distributed, real-time application. Our modeling and evaluation effort addresses two objectives: (1) providing early feedback to the system developers regarding the JEWEL IS configuration options for this application; and (2) evaluation of the design alternatives for an adaptive controller to control the overhead and intrusion of the JEWEL IS to a real-time video conferencing application. For JEWEL IS design, we compare two data collection and forwarding policies (collect-and-forward and batch-and-forward). For the design of the adaptive controller, we compare two adaptation policies (static and dynamic adaptation) and two policies to schedule the implementation of the control decisions (distributed and centralized scheduling). Results reported in this paper indicate that the batch-and-forward policy for IS design static adaptation policy with distributed scheduling for the adaptive controller design meet the domain-specific requirements.

A compiler-based approach to design and engineering of complex real-time systems

We present a compiler-based approach to the design and engineering of complex real-time systems. The systems are built from computation, communication and I/O subsystems, based on different extant real-time models and schemes. We have defined a high-level system specification language, called Real-Time System Markup Language (RTSML), for integration of model-specific subsystems. We have developed a prototype extensible compiler from RTSML to a Constraint Logic Programming (CLP) language. Model-specific compiler modules can be added that generate CLP code for subsystems based on the corresponding real-time models and schemes, and their integration with other subsystems of a complex real-time system being specified. We specify an example system consisting of dozens of communicating real-time tasks that share real-time computation and communication resources, and map it to two problem solving approaches. The first one is the conventional CLP search, while the second one is suitable for system reconfiguration when system parameters change dynamically. We compare the two approaches and present their timings for the example system. Based on the results obtained, and potentials of the software technologies used in our compiler-based approach, it appears promising for large-scale complex real-time systems.

Vista: a framework for instrumentation system design for multidisciplinary applications

This paper presents a framework, called Vista, for designing instrumentation systems (ISs) for various applications that use distributed computing systems. We define an instrumentation system as a collection of modules and functions used for collecting data from concurrent processes executed on various nodes across the network. Based on our generic model of an IS, the Vista framework provides various abstract and base classes that help a user to quickly configure an IS for a particular application domain. Vista library is built on top of various operating system services to provide various convenience functions to collect and forward the instrumentation data. The design of this framework is hierarchical to address diverse user needs. We present the application of this framework to various practical data collection and management problems.

Performance optimization of distributed applications in an extensible, adaptive environment

Abstract Management of performance data is central to the optimization process for complex, high-performance parallel and distributed systems. Management involves data collection, storage, transfer, buffering, filtering, processing, rendering, and presentation by various instrumentation, visualization, and analysis tools. Tool integration technologies and environments play a significant role in making different tools easier to combine and use. We present the PGRT environment that we have developed for these purposes. We describe how domain-specific instrumentation and visualization can be developed in PGRT and present two examples. The PGRT infrastructure is designed to support advanced analysis for resource scheduling, application and instrumentation steering, and performance data mining.

Integrating On-Line Performance Visualization and Real-Time System Design

Emerging parallel or distributed, heterogeneous real-time computer systems with many disparate constraints and requirements would benefit from a unifying and comprehensive systems engineering support in the design, testing and deployment phases, which interfaces with a human at a very high level and efficiently handles the system complexity. We present an approach to integration of (1) a compiler- and Constraint Logic Programming-based approach to design and engineering of real-time systems, and (2) an on-line performance analysis and visualization technology. An example distributed real-time system is used to help describe the integrated approach. Details are presented about how the involved technologies are used to help system developers and users achieve better system performance through on-line repair and reconfiguration.

An integrated approach to real-time system design and on-line performance visualization with steering

Emerging parallel or distributed heterogeneous real-time computer systems with many disparate constraints and requirements would benefit from a unifying and comprehensive systems engineering support in the design, testing and deployment phases, which interfaces with a human at a very high level and efficiently handles the system complexity. We present an approach to integration of: a compiler and constraint logic programming-based approach to design and engineering of complex real-time systems; and an online performance analysis and visualization technology with steering. An example distributed real-time system is used to help describe the integrated approach. Details are presented about how the involved technologies are used to help system developers and users achieve better system performance through on-line repair and reconfiguration.