Skip to content

Alpha version of AsyncFlow

Choose a tag to compare

View all tags
@GioeleB00 GioeleB00 released this 17 Aug 17:35
· 12 commits to main since this release
v0.1.0-alpha
7969be6
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

AsyncFlow v0.1.0-alpha — Release Notes (pre-release)

Summary

First public alpha of AsyncFlow, a SimPy-based, event-loop-aware simulator for asynchronous distributed systems.
Model clients, servers, load balancers, and network edges; run reproducible simulations; and analyze latency, throughput, and per-server resource metrics with built-in plots.

Highlights

  • Event-loop model: explicit CPU steps (blocking), I/O waits (non-blocking), and RAM residency per request.
  • Topology graph: request generator → client → servers; optional load balancer (round-robin) → multi-server fan-out; configurable network edges (stochastic latency).
  • Workload: stochastic traffic via simple random-variable configs (Poisson defaults).

Metrics

  • Event: RqsClock (end-to-end latencies).
  • Sampled: ready_queue_len, event_loop_io_sleep, ram_in_use, edge_concurrent_connection.

Analyzer API (ResultsAnalyzer):

  • get_latency_stats(), get_throughput_series()
  • Plots:
    • plot_latency_distribution(ax), plot_throughput(ax)
    • per-server:
      • plot_single_server_ready_queue(ax, id)
      • plot_single_server_io_queue(ax, id)
      • plot_single_server_ram(ax, id)
    • compact: plot_base_dashboard(...)

Examples

  • YAML quickstart (single server).
  • Pythonic builder (single server; load balancer + two servers).
  • All figures saved next to the script for easy inspection.

Tooling & CI

  • One-shot setup scripts: scripts/dev_setup.(sh|ps1)
    (installs Poetry, creates .venv, installs dev deps, runs ruff/mypy/pytest).
  • Convenience scripts: quality_check, run_tests, run_sys_tests.
  • CI (GitHub Actions): lint + type-check + tests; system tests gate merges into main.

Compatibility

  • Python 3.12+ (Linux/macOS/Windows).

Install from PyPI:

pip install asyncflow-sim

Known limitations (alpha)

  • Simplified network (latency + optional drops; no bandwidth/payload yet).
  • One event loop per server (no multi-process/node).
  • Linear endpoint pipelines (no branching/fan-out inside an endpoint).
  • Stationary workload; very short spikes may be missed if sample_period_s is large.

Roadmap (abridged)

  • Monte Carlo engine (CI/CBs)
  • Stochastic service times per step
  • Expanded component library (cache/DB/APIGW)
  • Fault/event injection
  • Richer network (bandwidth, retries/timeouts)
  • Branching and fan-out
  • Basic autoscaling/backpressure

Notes

Pre-release: APIs and plot styles may change before 0.1.0.
Feedback and issues are very welcome—this release is meant to validate the API and UX before the stable cut.

Assets 2
Loading