microsim - Microservices Simulator

Used in https://github.com/PacktPublishing/Mastering-Distributed-Tracing/tree/master/Chapter12.

Published on Docker Hub: https://hub.docker.com/r/yurishkuro/microsim/tags

Usage

To see all command line options:

docker run yurishkuro/microsim -h

microsim uses OpenTelemetry SDK to export traces. By default it will try to send them to https://localhost:4318/v1/traces (with TLS enabled). This can be changed by setting environment variables supported by the SDK:

• OTEL_EXPORTER_OTLP_TRACES_ENDPOINT to point to a different host/port/endpoint.
• OTEL_EXPORTER_OTLP_ENDPOINT to point to a different host/port (will still use default /v1/traces endpoint).
• OTEL_EXPORTER_OTLP_INSECURE=true if you just want to switch default URL to http instead of https.

Note that when we run microsim as a container, the localhost refers to the container's inner network namespace, so it will not be able to reach the collector even if it's running on the same host. You might see an error like this:

2024/07/02 18:06:07 traces export: Post "https://localhost:4318/v1/traces": dial tcp [::1]:4318: connect: connection refused

To work around that, refer to the IP address of the host instead:

docker run --env OTEL_EXPORTER_OTLP_ENDPOINT=http://{YOUR_IP_ADDRESS}:4318/ \
  yurishkuro/microsim -w=1 -r=1

Design

The tool takes a configuration file that describes the "architecture" of your desired system, such as services, number of instances for each service, their endpoints and their dependecies. The tool simulates the traces that may be generated from such system. Some random mutations are introduced during execution: latency of requests is drawn randomly from a normal distribution with specified mean and stdev (error rates can be specified, but currently not implemented).

The tool starts a configured number of "workers" (goroutines), and each worker sequentially runs simulations of requests through the architecture, starting from the root service. Workers can run either for a certain time period, or until they generate a predefined number of simulations. There is a configurable sleep interval between simulations in each worker.

Configuration

The tool comes with a built-in configuration for Jaeger's HotROD application that can be printed with:

docker run yurishkuro/microsim -o | jq

• A custom configuration can be provided via -c option.
• The schema for the config is hardcoded in the data model: model/config.go.
• The UML diagram below is generated with https://www.dumels.com/.

License

MIT license.