Skip to main content

Why not to use StopWatch when you need to measure the duration of an HTTP request in WebAPI

In this post we will talk about how we can measure how long it takes for a HTTP request to be executed on an ASP.NET MVC application.
All the tests are done using a web site hosted on Microsoft Azure. The instance used for this purpose is Shared - F1.

Let's assume that we have the following requirement:
At the end of each HTTP request you need to add to the logs information related to request duration.
The first solution that could come into our mind is to use "HttpContext.Current.Timestamp" to calculate the duration of a request. In theory we could calculate the difference between "DateTime.Now" and timestamp from "HttpContext".

protected void Application_EndRequest()
{
    Trace.WriteLine(string.Format("Request duration: {0}",
         (DateTime.Now - HttpContext.Current.Timestamp).TotalMilliseconds));
}

As we can see in the above example, we added this logic in the "Global.asax" file, in the "Application_EndRequest" method. Don't forget that the time format of HttpContext is local time not UTC time.

Out of topic: The code can be added in different locations, from "ActionFilter" to a HTTP Module. The location it is not important for now.

This solution is simple and nice, but the accurate of the duration is not the best one, but it is enough for most cases.
Remarks: Don't forget that "DateTime.Now" has a resolution accuracy of ~10ms.
But, the results are very good.
where:
  • min, avg and max is in ms
  • first table describes the status code of the http request
  • second table group all results based on request duration (first columns is request duration, second column contains the number of requests) 
For this, you could try to use Stopwatch and come with the following solution:
 
private static readonly ConcurrentDictionary<HttpRequest, Stopwatch> timeDictionary = 
     new ConcurrentDictionary<HttpRequest, Stopwatch>();

protected void Application_BeginRequest(object sender, EventArgs e)
{
    timeDictionary.TryAdd(
        Request, 
        Stopwatch.StartNew());
}

protected void Application_EndRequest(object sender, EventArgs e)
{
    Stopwatch sw = null;
    if (timeDictionary.TryGetValue(Request, out sw))
    {
        Trace.WriteLine(string.Format(
            "Request duration: {0}",
            sw.Elapsed));
    }
}

The above solution is using a Stopwatch to calculate the duration of each request. It will work great until ... you will hit 100 or 1.000 request per second. In that moment, if you have for example a Web API with a simple GET, like in the below example you will have 25s latency and a lot of timeouts.

[HttpGet]
public async Task<IHttpActionResult> GetFooAsync(string id)
{
    return Ok();
}

Results for 1.000 requests simultaneous:

You are wondering what are the results when you remove the logic from Begin and End requests. Let's take a look.
As we can see, without stopwatch we are pretty okay.
In comparison with the first test, that used "HttpContext.Current.Timestamp" the results are similar - +/- 1s is acceptable because the client was on my machine and the quality of internet connection can vary.

Le'ts try to not store the Stopwatch in a collection. We could add Stopwatch directly to the HttpRequest.

protected void Application_BeginRequest(object sender, EventArgs e)
{
    HttpApplication httpApp = (HttpApplication)sender;
    httpApp.Context.Items["Timer"] = Stopwatch.StartNew();
    ...
}


protected void Application_EndRequest(object sender, EventArgs e)
{
    var httpApp = (HttpApplication)sender;
    var sw = (Stopwatch)httpApp.Context.Items["Timer"];
    ...
}

The results are very similar with the one that we obtained using a collection of Stopwatch's.

  • min: 107ms     
  • avg: 25098.356ms     
  • max: 50489
The best solution
Of course the best solution is to use Performance Counters or a profiling tool like Glimse or similar tools. 

In conclusion we can say that using Stopwatch is a big mistake when we need to measure how long it takes for different block of code to execute when the load on the system is very high, especially on web application where we can use HttpContext.

Comments

Popular posts from this blog

Windows Docker Containers can make WIN32 API calls, use COM and ASP.NET WebForms

After the last post , I received two interesting questions related to Docker and Windows. People were interested if we do Win32 API calls from a Docker container and if there is support for COM. WIN32 Support To test calls to WIN32 API, let’s try to populate SYSTEM_INFO class. [StructLayout(LayoutKind.Sequential)] public struct SYSTEM_INFO { public uint dwOemId; public uint dwPageSize; public uint lpMinimumApplicationAddress; public uint lpMaximumApplicationAddress; public uint dwActiveProcessorMask; public uint dwNumberOfProcessors; public uint dwProcessorType; public uint dwAllocationGranularity; public uint dwProcessorLevel; public uint dwProcessorRevision; } ... [DllImport("kernel32")] static extern void GetSystemInfo(ref SYSTEM_INFO pSI); ... SYSTEM_INFO pSI = new SYSTEM_INFO(

ADO.NET provider with invariant name 'System.Data.SqlClient' could not be loaded

Today blog post will be started with the following error when running DB tests on the CI machine: threw exception: System.InvalidOperationException: The Entity Framework provider type 'System.Data.Entity.SqlServer.SqlProviderServices, EntityFramework.SqlServer' registered in the application config file for the ADO.NET provider with invariant name 'System.Data.SqlClient' could not be loaded. Make sure that the assembly-qualified name is used and that the assembly is available to the running application. See http://go.microsoft.com/fwlink/?LinkId=260882 for more information. at System.Data.Entity.Infrastructure.DependencyResolution.ProviderServicesFactory.GetInstance(String providerTypeName, String providerInvariantName) This error happened only on the Continuous Integration machine. On the devs machines, everything has fine. The classic problem – on my machine it’s working. The CI has the following configuration: TeamCity .NET 4.51 EF 6.0.2 VS2013 It see

Navigating Cloud Strategy after Azure Central US Region Outage

 Looking back, July 19, 2024, was challenging for customers using Microsoft Azure or Windows machines. Two major outages affected customers using CrowdStrike Falcon or Microsoft Azure computation resources in the Central US. These two outages affected many people and put many businesses on pause for a few hours or even days. The overlap of these two issues was a nightmare for travellers. In addition to blue screens in the airport terminals, they could not get additional information from the airport website, airline personnel, or the support line because they were affected by the outage in the Central US region or the CrowdStrike outage.   But what happened in reality? A faulty CrowdStrike update affected Windows computers globally, from airports and healthcare to small businesses, affecting over 8.5m computers. Even if the Falson Sensor software defect was identified and a fix deployed shortly after, the recovery took longer. In parallel with CrowdStrike, Microsoft provided a too