Practical JMT Tutorials: From Model Creation to Result Interpretation

Comparing JMT with Other Java Performance Modelling Tools

Overview

JMT (Java Modelling Tools) is a suite for building and analyzing queueing network models, focusing on performance evaluation of distributed and concurrent systems. When comparing JMT to other Java-based performance modelling tools, consider these dimensions: modeling approach, supported model types, usability, analysis methods, accuracy, extensibility, and community/support.

Key comparison points

• Modeling approach

  • JMT: Discrete-event and queueing network models with both open and closed networks; supports layered models and general service time distributions.
  • Others (e.g., Java Modelling Frameworks or libraries): May focus on discrete-event simulation only (event schedulers) or provide higher-level modeling DSLs; some emphasize simulation rather than analytic queueing solutions.

• Supported model types

  • JMT: Multi-class queueing networks, Markovian and non-Markovian service times, fork-join, finite population, and layered queueing.
  • Others: Tools like SimJava/JSIM emphasize general-purpose discrete-event simulation; frameworks such as LQN-based tools or QNAP2 focus on task-based or specific analytic models rather than full queueing-network breadth.

• Analysis methods

  • JMT: Offers both numerical (approximate/iterative solvers) and simulation-based analysis; includes mean-value analysis and product-form approximations where applicable.
  • Others: Some offer only simulation, others only analytic solvers; hybrid tools vary in solver robustness and convergence behavior.

• Usability & interface

  • JMT: Provides GUI model editors, XML model formats, and command-line utilities; moderate learning curve for complex models.
  • Others: Simulation libraries (SimJava) require programming; commercial or research tools may offer richer GUIs or integrated profilers.

• Extensibility & integration

  • JMT: Open-source, modular, with components that can be extended or scripted; integrates with data export and external tools.
  • Others: Libraries offer deep programmatic extensibility; commercial tools may provide APIs or plugins.

• Accuracy & performance

  • JMT: Strong for queueing-network scenarios; accuracy depends on chosen solver and assumptions (approximation vs. full simulation).
  • Others: Pure simulators can model arbitrary behaviors with high fidelity at cost of runtime; analytic-only tools may be faster but limited by modeling assumptions.

• Community & support

  • JMT: Academic origins, used in research—documentation and examples available; community size moderate.
  • Others: Commercial tools have vendor support; open-source simulators vary in community activity.

When to choose JMT

• You need specialized queueing network analysis (multi-class, layered models).
• You want a mix of analytic solvers and simulation within one suite.
• You prefer an open-source, extensible tool focused on performance evaluation.

Alternatives to consider

• SimJava / JSIM — general discrete-event simulation in Java.
• LQN tools / Layered Queueing Network-specific tools — task-centered performance modeling.
• Commercial performance modeling suites — for integrated GUIs, support, and enterprise features.
• Generic simulation frameworks (e.g., AnyLogic, SimPy via interop) — for broader system modeling needs.

Practical tip

Use JMT when your problem maps well to queueing-network semantics and you need both analytic and simulation-backed results; choose a pure simulator if you require modeling of complex, non-queueing behaviors or detailed system-level interactions.