Nlog vs Serilog with BenchmarkDotNet

This project benchmark Nlog and Serilog in a simple batch project in .Net6.

My goal is to find the best option for my program and also to provide you with an example for your own performance testing with Serilog/Nlog or others.

I use BenchmarkDotNet.

The tests

3 series of tests is used, sync and async. More later. - Serilog -Nlog -Nlog version 5.0 cr1

For a total of 6 tests.

Method
Serilog_DefaultExecution
Nlog_DefaultExecution
Nlog_Async
Serilog_Async
Nlog_5_0_Beta_DefaultExecution
Nlog_5_0_Beta_Async

Benchmark configuration

The Benchmark.csproj is the project where the benchmark tests is configure and run.

In the class BenchmarkCases.cs you can find the configuration of all tests. Each run all log file are deleted with the Cleanup() method and all test is run by the attribute [Benchmark()].

 [GlobalCleanup]
    public void Cleanup()
    {
        Directory.Delete(PATH, true);
    }

    [Benchmark(Baseline = true)]
    public void Nlog_DefaultExecution()
    {
        NlogProvider.Program.Main(new string[] { NB_EXECUTION, False, PATH });
    }

    [Benchmark()]
    public void Nlog_Async()
    {
        NlogProvider.Program.Main(new string[] { NB_EXECUTION, True, PATH });
    }

    [Benchmark()]
    public void Serilog_DefaultExecution()
    {
        SerilogProvider.Program.Main(new string[] { NB_EXECUTION, False, PATH });
    }

    [Benchmark()]
    public void Serilog_Async()
    {
        SerilogProvider.Program.Main(new string[] { NB_EXECUTION, True, PATH });
    }

How to run the benchmark

You need to run the Benchmark.csproj in Release mode and to start, go to debug/Start without debugging or Ctrl+F5. A Console will start de benchmark. it will take a few minutes.

Projects

Logger

Logger Is the batch exemple. It ILogger to write information log with Microsoft.Extensions.Logging

    using (_logger.BeginScope("DoSomethingElse"))
    {
        for (int i = 0; i < nbExecution; i++)
        {
            using (_logger.BeginScope("loop : {i}", i))
            {
                _logger.LogInformation("Begin loop : {i}", i);

                Thread.Sleep(100);

                _logger.LogInformation("End loop : {i}", i);
            }
        }

        _logger.LogInformation("End DoSomethingElse");
    }

NlogProvider

This project uses NLog provider to log into a file the log of Logger project.

 .ConfigureLogging((context, logging) =>
    {
        logging.ClearProviders();
        logging.SetMinimumLevel(Microsoft.Extensions.Logging.LogLevel.Debug);
        SetupLogger(Path.Combine(path, "Nlog.log"));
    })
    .UseNLog();

SerilogProvider

This project uses Serilog provider to log into a file the log of Logger project.

host.UseSerilog((context, logger) =>
{
    logger.Enrich.FromLogContext();
    logger.WriteTo.File(fileName,
        outputTemplate: GetOutputTemplate()
        );
});

The result

My pc setup

Infos
Nom du système d’exploitation:	Microsoft Windows 11 Professionnel
Processeur:	Intel(R) Core(TM) i7-8700K CPU @ 3.70GHz, 3696 MHz, 6 cœur(s), 12 processeur(s) logique(s)
Mémoire physique totale:	32 701 Mo
Description de la carte:	NVIDIA GeForce GTX 1080 Ti

BenchmarkDotNet=v0.13.1, OS=Windows 10.0.22000
Intel Core i7-8700K CPU 3.70GHz (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET SDK=6.0.101
  [Host]     : .NET 6.0.1 (6.0.121.56705), X64 RyuJIT
  DefaultJob : .NET 6.0.1 (6.0.121.56705), X64 RyuJIT

Result with thread.sleep(100)

Result Folder : NlogSerilogDotnetBenchmark\Benchmark\bin\Release\net6.0\BenchmarkDotNet.Artifacts\results

Method	Mean	Error	StdDev	Min	Max	Ratio	Rank	Allocated
Serilog_DefaultExecution	5.437 s	0.0081 s	0.0076 s	5.423 s	5.448 s	1.00	1	341 KB
Nlog_5_0_Beta_Async	5.438 s	0.0078 s	0.0073 s	5.425 s	5.448 s	1.00	1	354 KB
Serilog_Async	5.438 s	0.0078 s	0.0073 s	5.428 s	5.451 s	1.00	1	344 KB
Nlog_DefaultExecution	5.439 s	0.0081 s	0.0068 s	5.428 s	5.450 s	1.00	1	370 KB
Nlog_Async	5.439 s	0.0097 s	0.0090 s	5.426 s	5.457 s	1.00	1	351 KB
Nlog_5_0_Beta_DefaultExecution	5.444 s	0.0094 s	0.0088 s	5.434 s	5.458 s	1.00	1	345 KB

Result without thread.sleep(100)

Because the execution time was long, I reduced the loop to 10 instead of 50. The results is very different !

Method	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD	Rank	Gen 0	Gen 1	Allocated
Serilog_DefaultExecution	563.5 μs	45.52 μs	126.90 μs	495.6 μs	483.0 μs	970.2 μs	0.36	0.10	1	25.3906	8.7891	154 KB
Nlog_5_0_Beta_Async	1,284.5 μs	24.97 μs	23.36 μs	1,286.9 μs	1,239.7 μs	1,320.5 μs	0.99	0.05	2	41.0156	35.1563	250 KB
Nlog_DefaultExecution	1,676.1 μs	102.06 μs	300.94 μs	1,626.5 μs	1,209.7 μs	2,310.5 μs	1.00	0.00	3	39.0625	9.7656	243 KB
Nlog_5_0_Beta_DefaultExecution	1,687.3 μs	108.87 μs	321.01 μs	1,649.7 μs	1,195.6 μs	2,375.9 μs	1.01	0.05	3	39.0625	29.2969	244 KB
Nlog_Async	1,711.7 μs	106.24 μs	313.25 μs	1,690.5 μs	1,261.3 μs	2,486.2 μs	1.02	0.04	3	41.0156	9.7656	250 KB
Serilog_Async	9,761.5 μs	1,262.12 μs	3,721.40 μs	10,280.2 μs	3,661.7 μs	17,371.4 μs	5.62	1.33	4	25.3906	9.7656	157 KB

Manual test (Elapsed time)

With manual testing that speaks to me a bit more, don't you?

Provider	100	500	1000	10 000	100 000
Nlog default	471 ms	717 ms	648 ms	625 ms	1878 ms
Nlog Async	18 ms	38 ms	43 ms	222 ms	1012 ms
Nlog_5.0 default	8 ms	24 ms	36 ms	228 ms	978 ms
Nlog_5.0 Async	4 ms	17 ms	21 ms	249 ms	984 ms
Serilog default	145 ms	96 ms	125 ms	464 ms	2126 ms
Serilog Async	14 ms	22 ms	51 ms	160 ms	991 ms

Conclusion

In manual tests, Nlog 5.0 and Serilog Async look great on each try. Nlog Async is not too bad to.

With benchmark the rank is different, Serilog with default setting seem the best but in elapsed time (manual test) he is not.

I hope this example will help you in your process of finding the one that is best for your project.

Thanks

jakdaniel