Installation
This page installs desflow and runs a first simulation that needs no containers.
Install the Package
Install the core library with uv or pip:
desflow was called dynamic-des before 0.20.0. The package, the Python module and the command are now all named desflow. Version 1.0 removed the dynamic_des module and the ddes command. Code that imports dynamic_des and scripts that run ddes must change to desflow, and Upgrading to 1.0 says what to change.
Connectors and input fitting are optional extras. For example, uv add "desflow[kafka,parquet]" or pip install "desflow[kafka,parquet]", and uv tool install "desflow[kafka,parquet]" for the desflow command:
| Extra | Adds |
|---|---|
kafka |
Kafka ingress and egress |
confluent |
Avro, JSON Schema and Protobuf with the Confluent Schema Registry (includes kafka) |
glue |
Avro, JSON Schema and Protobuf with the AWS Glue Schema Registry (includes kafka) |
redis |
Redis ingress and egress |
postgres |
PostgreSQL ingress and egress |
parquet |
Parquet and JSONL files on local disk, S3-compatible storage, Google Cloud Storage or Azure Blob Storage |
iceberg |
Apache Iceberg tables through a REST catalog, a SQL catalog on SQLite or PostgreSQL, or the Glue catalog |
fit |
scipy and matplotlib, for input fitting with desflow fit |
fake |
Faker, for the fake generator of realistic names, addresses and dates |
watch |
watchfiles, for desflow run --watch, which runs the file again each time it changes |
all |
Every extra above |
Two cautions apply to the fake extra. A seeded fake field gives the same values again only with the same version of Faker, because 1.0 does not record the version. A Faker method that reads the date, such as date_between, reads the wall clock and not the simulation clock, so it gives other values on another day.
On Python 3.15, the kafka, confluent and glue extras build aiokafka, and the iceberg extra builds pyiceberg and zstandard, from source, because those packages publish no wheel for 3.15 yet. A build from source needs a C compiler.
Run desflow on its own to list its commands. Command Line uses each command on one shop, the model of the experiment example, and links the full help of each.
Run an Example
The local example models a shop that receives and packs customer orders. The simulation prints events and telemetry to the terminal. The example is written three ways: as a YAML blueprint, with the builder and with the low-level API. Each version runs in real time and ends on its own after 30 seconds. Pick the way you write models, then follow uv or pip:
Quick Start
desflow examples get downloads the example at the release you install, into a folder of its own. Pick uv or pip:
Quick Start
desflow examples get downloads the example at the release you install, into a folder of its own. Pick uv or pip:
Quick Start
desflow examples get downloads the example at the release you install, into a folder of its own. Pick uv or pip:
desflow examples lists every example. Examples that need Kafka, a database or an object store start it with odctl, and each example page shows how.
Next Steps
- Learn: teaches discrete-event simulation and desflow from the start, through a tutorial that builds one shop model. Follow the tutorial in order. Read the Simulation Concepts pages for the theory after the tutorial, or when a tutorial page links to one.
- User Guide: describes each part of desflow, what it does and all its options. Read a page when you need that part.
- How-To: does one task per page, step by step, from modelling to connectors and analysis. It also holds complete examples to download and run.