- Originally aimed for developers (and up to corner case scenarios) for maximum flexibility.
- But easy enough to use it as writing integration tests - just for more threads. :)
- NuGet:
Install-Package Viki.LoadRunner -pre
Current documentation is only in form of code examples.
- It's far from completed, but IMO its should be good enough to see whether this tool can be useful.
Start with this small setup to get a feel on how it's configured.
- One needs to define 3 things to make fully working test (See QuickStartDemo.cs):
- [1] Write Scenario implementation, defining code it-self which will get executed concurrently multiple times.
- [2][Optional] Configure how results get aggregated/presented.
- [3] Configure strategy on how test gets executed (e.g. How much threads, how long it should run. etc...)
- Real case & simple load-test scenario to meassure "somewhat-theoretical" performance of simple http server - https://github.com/Vycka/HttpMockSlim/tree/master/tests/HttpMockSlim.LoadTest
HistogramAggregator is a default tool to aggregate results with defined dimensions/metrics
- Given its flexibility documentation will take time till its done.
- Until then - it should be easy enough to figure it out through various uses in examples already available:
- There is also a generic version of this Histogram to use it on your own custom data:
new Histogram<T>()- It comes with few generic dimensions and metrics, but given the custom type and required aggregation, one might need to implement some custom IDimension<T>'s and IMetric<T>'s
Rest of the demo project:
- Index
- Feature specific demos (WiP)
- Engine throughut (WiP)
- TheoreticalSpeedDemo.cs - Theoretical throughput without doing any measurements.
- AggregationImpactDemo.cs - Theoretical throughput with aggregation pipeline attached.
Until demo project is completed, one can also checkout my messy setups i use for debugging:
Have any questions/suggestions/etc, feel free to contact me.