pareto-simulator
Version:
Simulates request times following a Pareto distribution
55 lines (38 loc) • 1.5 kB
Markdown
Simulates a number of incoming requests with response times
following a [Pareto distribution](https://en.wikipedia.org/wiki/Pareto_distribution):
a very simple power law distribution.
Run from your console:
```
$ npm install -g pareto-simulator
```
Now you can run the Pareto simulator as a command:
```
$ pareto --xm 28
```
To simulate a search on the
[](https://www.computerweekly.com/news/2240088495/Single-Google-search-uses-1000-servers):
```
$ pareto --xm 1 -n 1000 --parallel 30 --series 30 --timeout 10 --linear
```
This simulates a request that branches out to 30 * 30 servers.
There are 30 steps, each consisting of 30 parallel invocations.
Each service is designed to take at least 1 millisecond,
with response times following a Pareto distribution,
and with a timeout of 10 ms.
The result is a nice linear graph that approaches a normal distribution.
Use `--help` to see all the options.
```
The following options are supported:
-a, --alpha <ARG1> Alpha parameter for Pareto ("1.16" by default)
-x, --xm <ARG1> Xm (minimum) for Pareto (required)
-n, --number <ARG1> Number of requests to simulate ("100000" by default)
-t, --timeout <ARG1> Timeout
-p, --parallel <ARG1> Requests in parallel ("1" by default)
-s, --series <ARG1> Consecutive requests ("1" by default)
```
## Acknowledgements
(C) Alex Fernández. Published under the [MIT license](LICENSE).