Event Service Simulation Animator and Debugger

Project SIENA

Principal Investigators: David S. Rosenblum (UC Irvine) and Alexander L. Wolf (University of Colorado at Boulder)

Customer Contact: Antonio Carzaniga (University of Colorado at Boulder)

Overview and Motivation

A common architectural style for distributed, loosely-coupled, heterogeneous software systems is a structure based on event generation, observation and notification. A notable characteristic of an architecture based on events is that interaction among architectural components occurs asynchronously, thereby simplifying the composition of autonomous, independently-executing components that may be written in different programming languages and executing on varied hardware platforms.

There is increasing interest in deploying these kinds of distributed systems across wide-area networks such as the Internet. For instance, workflow systems for multi-national corporations, multi-site/multi-organization software development, and real-time investment analysis across world financial markets are just a few of the many applications that lend themselves to deployment on an Internet scale. However, deployment of such systems at the scale of the Internet imposes new challenges that are not met by existing technology.

Objective

We have begun to design and implement a wide-area event observation and notification facility. It is clear, based on our experience, that simply trying to scale a design or an existing system that is intended for a local-area network is a flawed approach. We are therefore formulating new architectures and related algorithms for event notification that are scalable to the Internet. To aid this formulation, we are building a simulation tool that will allow us to analyze and adapt known Internet-scale architectures, such as that of Network News and the Domain Name Service. The tool allows us to simulate event observation and notification performance behavior in a variety of wide-area network scenarios. The evaluation, analysis and debugging of simulated architectures and algorithms would be improved greatly by having a visualization tool for the simulator that aids animation and debugging of simulation traces. The objective of this project is therefore to build such a visualization tool.

Required Components

The animator/debugger would contain the following components:

• Input processing: An animation is driven by four kinds of input files: (1) a topology specification for a particular simulation run, (2) an optional process deployment specification specifying how objects are deployed across the simulated topology, (3) a mapping specification of event service site names to site IDs, and (4) a simulation trace.
• Simulation animator: a graphical animation of the simulation trace that displays the network topology, with simulation behavior highlighted as the trace is "executed".
• Animation options: The animation would be highly configurable by a number of layout, playback, debugging, view filtering and clock manipulation options.

References

Additional details are available from the SIENA project personnel. An overview of the work is presented in the following paper:

David S. Rosenblum and Alexander L. Wolf, A Design Framework for Internet-Scale Event Observation and Notification, Proc. Sixth European Software Engineering Conf./ACM SIGSOFT Fifth Symposium on the Foundations of Software Engineering, Zurich, Switzerland, Sep. 1997, pp. 344-360.