Muokkaa

Jaa

Rate limiting middleware in ASP.NET Core

  • Artikkeli

By Arvin Kahbazi, Maarten Balliauw, and Rick Anderson

The Microsoft.AspNetCore.RateLimiting middleware provides rate limiting middleware. Apps configure rate limiting policies and then attach the policies to endpoints. Apps using rate limiting should be carefully load tested and reviewed before deploying. See Testing endpoints with rate limiting in this article for more information.

For an introduction to rate limiting, see Rate limiting middleware.

Rate limiter algorithms

The RateLimiterOptionsExtensions class provides the following extension methods for rate limiting:

Fixed window limiter

The AddFixedWindowLimiter method uses a fixed time window to limit requests. When the time window expires, a new time window starts and the request limit is reset.

Consider the following code:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddRateLimiter(_ => _
    .AddFixedWindowLimiter(policyName: "fixed", options =>
    {
        options.PermitLimit = 4;
        options.Window = TimeSpan.FromSeconds(12);
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = 2;
    }));

var app = builder.Build();

app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", () => Results.Ok($"Hello {GetTicks()}"))
                           .RequireRateLimiting("fixed");

app.Run();

The preceding code:

Apps should use Configuration to set limiter options. The following code updates the preceding code using MyRateLimitOptions for configuration:

using System.Threading.RateLimiting;
using Microsoft.AspNetCore.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);
builder.Services.Configure<MyRateLimitOptions>(
    builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit));

var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);
var fixedPolicy = "fixed";

builder.Services.AddRateLimiter(_ => _
    .AddFixedWindowLimiter(policyName: fixedPolicy, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var app = builder.Build();

app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", () => Results.Ok($"Fixed Window Limiter {GetTicks()}"))
                           .RequireRateLimiting(fixedPolicy);

app.Run();

UseRateLimiter must be called after UseRouting when rate limiting endpoint specific APIs are used. For example, if the [EnableRateLimiting] attribute is used, UseRateLimiter must be called after UseRouting. When calling only global limiters, UseRateLimiter can be called before UseRouting.

Sliding window limiter

A sliding window algorithm:

  • Is similar to the fixed window limiter but adds segments per window. The window slides one segment each segment interval. The segment interval is (window time)/(segments per window).
  • Limits the requests for a window to permitLimit requests.
  • Each time window is divided in n segments per window.
  • Requests taken from the expired time segment one window back (n segments prior to the current segment) are added to the current segment. We refer to the most expired time segment one window back as the expired segment.

Consider the following table that shows a sliding window limiter with a 30-second window, three segments per window, and a limit of 100 requests:

  • The top row and first column shows the time segment.
  • The second row shows the remaining requests available. The remaining requests are calculated as the available requests minus the processed requests plus the recycled requests.
  • Requests at each time moves along the diagonal blue line.
  • From time 30 on, the request taken from the expired time segment are added back to the request limit, as shown in the red lines.

Table showing requests, limits, and recycled slots

The following table shows the data in the previous graph in a different format. The Available column shows the requests available from the previous segment (The Carry over from the previous row). The first row shows 100 available requests because there's no previous segment.

Time Available Taken Recycled from expired Carry over
0 100 20 0 80
10 80 30 0 50
20 50 40 0 10
30 10 30 20 0
40 0 10 30 20
50 20 10 40 50
60 50 35 30 45

The following code uses the sliding window rate limiter:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);
var slidingPolicy = "sliding";

builder.Services.AddRateLimiter(_ => _
    .AddSlidingWindowLimiter(policyName: slidingPolicy, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.SegmentsPerWindow = myOptions.SegmentsPerWindow;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var app = builder.Build();

app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", () => Results.Ok($"Sliding Window Limiter {GetTicks()}"))
                           .RequireRateLimiting(slidingPolicy);

app.Run();

Token bucket limiter

The token bucket limiter is similar to the sliding window limiter, but rather than adding back the requests taken from the expired segment, a fixed number of tokens are added each replenishment period. The tokens added each segment can't increase the available tokens to a number higher than the token bucket limit. The following table shows a token bucket limiter with a limit of 100 tokens and a 10-second replenishment period.

Time Available Taken Added Carry over
0 100 20 0 80
10 80 10 20 90
20 90 5 15 100
30 100 30 20 90
40 90 6 16 100
50 100 40 20 80
60 80 50 20 50

The following code uses the token bucket limiter:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

var tokenPolicy = "token";
var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);

builder.Services.AddRateLimiter(_ => _
    .AddTokenBucketLimiter(policyName: tokenPolicy, options =>
    {
        options.TokenLimit = myOptions.TokenLimit;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
        options.ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod);
        options.TokensPerPeriod = myOptions.TokensPerPeriod;
        options.AutoReplenishment = myOptions.AutoReplenishment;
    }));

var app = builder.Build();

app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", () => Results.Ok($"Token Limiter {GetTicks()}"))
                           .RequireRateLimiting(tokenPolicy);

app.Run();

When AutoReplenishment is set to true, an internal timer replenishes the tokens every ReplenishmentPeriod; when set to false, the app must call TryReplenish on the limiter.

Concurrency limiter

The concurrency limiter limits the number of concurrent requests. Each request reduces the concurrency limit by one. When a request completes, the limit is increased by one. Unlike the other requests limiters that limit the total number of requests for a specified period, the concurrency limiter limits only the number of concurrent requests and doesn't cap the number of requests in a time period.

The following code uses the concurrency limiter:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

var concurrencyPolicy = "Concurrency";
var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);

builder.Services.AddRateLimiter(_ => _
    .AddConcurrencyLimiter(policyName: concurrencyPolicy, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var app = builder.Build();

app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", async () =>
{
    await Task.Delay(500);
    return Results.Ok($"Concurrency Limiter {GetTicks()}");
                              
}).RequireRateLimiting(concurrencyPolicy);

app.Run();

Create chained limiters

The CreateChained API allows passing in multiple PartitionedRateLimiter which are combined into one PartitionedRateLimiter. The combined limiter runs all the input limiters in sequence.

The following code uses CreateChained:

using System.Globalization;
using System.Threading.RateLimiting;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddRateLimiter(_ =>
{
    _.OnRejected = (context, _) =>
    {
        if (context.Lease.TryGetMetadata(MetadataName.RetryAfter, out var retryAfter))
        {
            context.HttpContext.Response.Headers.RetryAfter =
                ((int) retryAfter.TotalSeconds).ToString(NumberFormatInfo.InvariantInfo);
        }

        context.HttpContext.Response.StatusCode = StatusCodes.Status429TooManyRequests;
        context.HttpContext.Response.WriteAsync("Too many requests. Please try again later.");

        return new ValueTask();
    };
    _.GlobalLimiter = PartitionedRateLimiter.CreateChained(
        PartitionedRateLimiter.Create<HttpContext, string>(httpContext =>
        {
            var userAgent = httpContext.Request.Headers.UserAgent.ToString();

            return RateLimitPartition.GetFixedWindowLimiter
            (userAgent, _ =>
                new FixedWindowRateLimiterOptions
                {
                    AutoReplenishment = true,
                    PermitLimit = 4,
                    Window = TimeSpan.FromSeconds(2)
                });
        }),
        PartitionedRateLimiter.Create<HttpContext, string>(httpContext =>
        {
            var userAgent = httpContext.Request.Headers.UserAgent.ToString();
            
            return RateLimitPartition.GetFixedWindowLimiter
            (userAgent, _ =>
                new FixedWindowRateLimiterOptions
                {
                    AutoReplenishment = true,
                    PermitLimit = 20,    
                    Window = TimeSpan.FromSeconds(30)
                });
        }));
});

var app = builder.Build();
app.UseRateLimiter();

static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/", () => Results.Ok($"Hello {GetTicks()}"));

app.Run();

For more information, see the CreateChained source code

EnableRateLimiting and DisableRateLimiting attributes

The [EnableRateLimiting] and [DisableRateLimiting] attributes can be applied to a Controller, action method, or Razor Page. For Razor Pages, the attribute must be applied to the Razor Page and not the page handlers. For example, [EnableRateLimiting] can't be applied to OnGet, OnPost, or any other page handler.

The [DisableRateLimiting] attribute disables rate limiting to the Controller, action method, or Razor Page regardless of named rate limiters or global limiters applied. For example, consider the following code which calls RequireRateLimiting to apply the fixedPolicy rate limiting to all controller endpoints:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddRazorPages();
builder.Services.AddControllersWithViews();

builder.Services.Configure<MyRateLimitOptions>(
    builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit));

var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);
var fixedPolicy = "fixed";

builder.Services.AddRateLimiter(_ => _
    .AddFixedWindowLimiter(policyName: fixedPolicy, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var slidingPolicy = "sliding";

builder.Services.AddRateLimiter(_ => _
    .AddSlidingWindowLimiter(policyName: slidingPolicy, options =>
    {
        options.PermitLimit = myOptions.SlidingPermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.SegmentsPerWindow = myOptions.SegmentsPerWindow;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var app = builder.Build();
app.UseRateLimiter();

if (!app.Environment.IsDevelopment())
{
    app.UseExceptionHandler("/Error");
    app.UseHsts();
}

app.UseHttpsRedirection();
app.UseStaticFiles();

app.MapRazorPages().RequireRateLimiting(slidingPolicy);
app.MapDefaultControllerRoute().RequireRateLimiting(fixedPolicy);

app.Run();

In the following code, [DisableRateLimiting] disables rate limiting and overrides [EnableRateLimiting("fixed")] applied to the Home2Controller and app.MapDefaultControllerRoute().RequireRateLimiting(fixedPolicy) called in Program.cs:

[EnableRateLimiting("fixed")]
public class Home2Controller : Controller
{
    private readonly ILogger<Home2Controller> _logger;

    public Home2Controller(ILogger<Home2Controller> logger)
    {
        _logger = logger;
    }

    public ActionResult Index()
    {
        return View();
    }

    [EnableRateLimiting("sliding")]
    public ActionResult Privacy()
    {
        return View();
    }

    [DisableRateLimiting]
    public ActionResult NoLimit()
    {
        return View();
    }

    [ResponseCache(Duration = 0, Location = ResponseCacheLocation.None, NoStore = true)]
    public IActionResult Error()
    {
        return View(new ErrorViewModel { RequestId = Activity.Current?.Id ?? HttpContext.TraceIdentifier });
    }
}

In the preceding code, the [EnableRateLimiting("sliding")] is not applied to the Privacy action method because Program.cs called app.MapDefaultControllerRoute().RequireRateLimiting(fixedPolicy).

Consider the following code which doesn't call RequireRateLimiting on MapRazorPages or MapDefaultControllerRoute:

using Microsoft.AspNetCore.RateLimiting;
using System.Threading.RateLimiting;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddRazorPages();
builder.Services.AddControllersWithViews();

builder.Services.Configure<MyRateLimitOptions>(
    builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit));

var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);
var fixedPolicy = "fixed";

builder.Services.AddRateLimiter(_ => _
    .AddFixedWindowLimiter(policyName: fixedPolicy, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var slidingPolicy = "sliding";

builder.Services.AddRateLimiter(_ => _
    .AddSlidingWindowLimiter(policyName: slidingPolicy, options =>
    {
        options.PermitLimit = myOptions.SlidingPermitLimit;
        options.Window = TimeSpan.FromSeconds(myOptions.Window);
        options.SegmentsPerWindow = myOptions.SegmentsPerWindow;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    }));

var app = builder.Build();

app.UseRateLimiter();

if (!app.Environment.IsDevelopment())
{
    app.UseExceptionHandler("/Error");
    app.UseHsts();
}

app.UseHttpsRedirection();
app.UseStaticFiles();

app.MapRazorPages();
app.MapDefaultControllerRoute();  // RequireRateLimiting not called

app.Run();

Consider the following controller:

[EnableRateLimiting("fixed")]
public class Home2Controller : Controller
{
    private readonly ILogger<Home2Controller> _logger;

    public Home2Controller(ILogger<Home2Controller> logger)
    {
        _logger = logger;
    }

    public ActionResult Index()
    {
        return View();
    }

    [EnableRateLimiting("sliding")]
    public ActionResult Privacy()
    {
        return View();
    }

    [DisableRateLimiting]
    public ActionResult NoLimit()
    {
        return View();
    }

    [ResponseCache(Duration = 0, Location = ResponseCacheLocation.None, NoStore = true)]
    public IActionResult Error()
    {
        return View(new ErrorViewModel { RequestId = Activity.Current?.Id ?? HttpContext.TraceIdentifier });
    }
}

In the preceding controller:

  • The "fixed" policy rate limiter is applied to all action methods that don't have EnableRateLimiting and DisableRateLimiting attributes.
  • The "sliding" policy rate limiter is applied to the Privacy action.
  • Rate limiting is disabled on the NoLimit action method.

Applying attributes to Razor Pages

For Razor Pages, the attribute must be applied to the Razor Page and not the page handlers. For example, [EnableRateLimiting] can't be applied to OnGet, OnPost, or any other page handler.

The DisableRateLimiting attribute disables rate limiting on a Razor Page. EnableRateLimiting is only applied to a Razor Page if MapRazorPages().RequireRateLimiting(Policy) has not been called.

Limiter algorithm comparison

The fixed, sliding, and token limiters all limit the maximum number of requests in a time period. The concurrency limiter limits only the number of concurrent requests and doesn't cap the number of requests in a time period. The cost of an endpoint should be considered when selecting a limiter. The cost of an endpoint includes the resources used, for example, time, data access, CPU, and I/O.

Rate limiter samples

The following samples aren't meant for production code but are examples on how to use the limiters.

Limiter with OnRejected, RetryAfter, and GlobalLimiter

The following sample:

  • Creates a RateLimiterOptions.OnRejected callback that is called when a request exceeds the specified limit. retryAfter can be used with the TokenBucketRateLimiter, FixedWindowLimiter, and SlidingWindowLimiter because these algorithms are able to estimate when more permits will be added. The ConcurrencyLimiter has no way of calculating when permits will be available.

  • Adds the following limiters:

    • A SampleRateLimiterPolicy which implements the IRateLimiterPolicy<TPartitionKey> interface. The SampleRateLimiterPolicy class is shown later in this article.
    • A SlidingWindowLimiter:
      • With a partition for each authenticated user.
      • One shared partition for all anonymous users.
    • A GlobalLimiter that is applied to all requests. The global limiter will be executed first, followed by the endpoint-specific limiter, if one exists. The GlobalLimiter creates a partition for each IPAddress.
// Preceding code removed for brevity.
using System.Globalization;
using System.Net;
using System.Threading.RateLimiting;
using Microsoft.AspNetCore.Identity;
using Microsoft.EntityFrameworkCore;
using WebRateLimitAuth;
using WebRateLimitAuth.Data;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

var connectionString = builder.Configuration.GetConnectionString("DefaultConnection") ??
    throw new InvalidOperationException("Connection string 'DefaultConnection' not found.");
builder.Services.AddDbContext<ApplicationDbContext>(options =>
    options.UseSqlServer(connectionString));
builder.Services.AddDatabaseDeveloperPageExceptionFilter();

builder.Services.AddDefaultIdentity<IdentityUser>(options => options.SignIn.RequireConfirmedAccount = true)
    .AddEntityFrameworkStores<ApplicationDbContext>();

builder.Services.Configure<MyRateLimitOptions>(
    builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit));

builder.Services.AddRazorPages();
builder.Services.AddControllersWithViews();

var userPolicyName = "user";
var helloPolicy = "hello";
var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);

builder.Services.AddRateLimiter(limiterOptions =>
{
    limiterOptions.OnRejected = (context, cancellationToken) =>
    {
        if (context.Lease.TryGetMetadata(MetadataName.RetryAfter, out var retryAfter))
        {
            context.HttpContext.Response.Headers.RetryAfter =
                ((int) retryAfter.TotalSeconds).ToString(NumberFormatInfo.InvariantInfo);
        }

        context.HttpContext.Response.StatusCode = StatusCodes.Status429TooManyRequests;
        context.HttpContext.RequestServices.GetService<ILoggerFactory>()?
            .CreateLogger("Microsoft.AspNetCore.RateLimitingMiddleware")
            .LogWarning("OnRejected: {GetUserEndPoint}", GetUserEndPoint(context.HttpContext));

        return new ValueTask();
    };

    limiterOptions.AddPolicy<string, SampleRateLimiterPolicy>(helloPolicy);
    limiterOptions.AddPolicy(userPolicyName, context =>
    {
        var username = "anonymous user";
        if (context.User.Identity?.IsAuthenticated is true)
        {
            username = context.User.ToString()!;
        }

        return RateLimitPartition.GetSlidingWindowLimiter(username,
            _ => new SlidingWindowRateLimiterOptions
            {
                PermitLimit = myOptions.PermitLimit,
                QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                QueueLimit = myOptions.QueueLimit,
                Window = TimeSpan.FromSeconds(myOptions.Window),
                SegmentsPerWindow = myOptions.SegmentsPerWindow
            });

    });
    
    limiterOptions.GlobalLimiter = PartitionedRateLimiter.Create<HttpContext, IPAddress>(context =>
    {
        IPAddress? remoteIpAddress = context.Connection.RemoteIpAddress;

        if (!IPAddress.IsLoopback(remoteIpAddress!))
        {
            return RateLimitPartition.GetTokenBucketLimiter
            (remoteIpAddress!, _ =>
                new TokenBucketRateLimiterOptions
                {
                    TokenLimit = myOptions.TokenLimit2,
                    QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                    QueueLimit = myOptions.QueueLimit,
                    ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod),
                    TokensPerPeriod = myOptions.TokensPerPeriod,
                    AutoReplenishment = myOptions.AutoReplenishment
                });
        }

        return RateLimitPartition.GetNoLimiter(IPAddress.Loopback);
    });
});

var app = builder.Build();

if (app.Environment.IsDevelopment())
{
    app.UseMigrationsEndPoint();
}
else
{
    app.UseExceptionHandler("/Error");
    app.UseHsts();
}

app.UseHttpsRedirection();
app.UseStaticFiles();

app.UseRouting();
app.UseRateLimiter();

app.UseAuthentication();
app.UseAuthorization();

app.MapRazorPages().RequireRateLimiting(userPolicyName);
app.MapDefaultControllerRoute();

static string GetUserEndPoint(HttpContext context) =>
   $"User {context.User.Identity?.Name ?? "Anonymous"} endpoint:{context.Request.Path}"
   + $" {context.Connection.RemoteIpAddress}";
static string GetTicks() => (DateTime.Now.Ticks & 0x11111).ToString("00000");

app.MapGet("/a", (HttpContext context) => $"{GetUserEndPoint(context)} {GetTicks()}")
    .RequireRateLimiting(userPolicyName);

app.MapGet("/b", (HttpContext context) => $"{GetUserEndPoint(context)} {GetTicks()}")
    .RequireRateLimiting(helloPolicy);

app.MapGet("/c", (HttpContext context) => $"{GetUserEndPoint(context)} {GetTicks()}");

app.Run();

Warning

Creating partitions on client IP addresses makes the app vulnerable to Denial of Service Attacks which employ IP Source Address Spoofing. For more information, see BCP 38 RFC 2827 Network Ingress Filtering: Defeating Denial of Service Attacks which employ IP Source Address Spoofing.

See the samples repository for the complete Program.cs file.

The SampleRateLimiterPolicy class

using System.Threading.RateLimiting;
using Microsoft.AspNetCore.RateLimiting;
using Microsoft.Extensions.Options;
using WebRateLimitAuth.Models;

namespace WebRateLimitAuth;

public class SampleRateLimiterPolicy : IRateLimiterPolicy<string>
{
    private Func<OnRejectedContext, CancellationToken, ValueTask>? _onRejected;
    private readonly MyRateLimitOptions _options;

    public SampleRateLimiterPolicy(ILogger<SampleRateLimiterPolicy> logger,
                                   IOptions<MyRateLimitOptions> options)
    {
        _onRejected = (ctx, token) =>
        {
            ctx.HttpContext.Response.StatusCode = StatusCodes.Status429TooManyRequests;
            logger.LogWarning($"Request rejected by {nameof(SampleRateLimiterPolicy)}");
            return ValueTask.CompletedTask;
        };
        _options = options.Value;
    }

    public Func<OnRejectedContext, CancellationToken, ValueTask>? OnRejected => _onRejected;

    public RateLimitPartition<string> GetPartition(HttpContext httpContext)
    {
        return RateLimitPartition.GetSlidingWindowLimiter(string.Empty,
            _ => new SlidingWindowRateLimiterOptions
            {
                PermitLimit = _options.PermitLimit,
                QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                QueueLimit = _options.QueueLimit,
                Window = TimeSpan.FromSeconds(_options.Window),
                SegmentsPerWindow = _options.SegmentsPerWindow
            });
    }
}

In the preceding code, OnRejected uses OnRejectedContext to set the response status to 429 Too Many Requests. The default rejected status is 503 Service Unavailable.

Limiter with authorization

The following sample uses JSON Web Tokens (JWT) and creates a partition with the JWT access token. In a production app, the JWT would typically be provided by a server acting as a Security token service (STS). For local development, the dotnet user-jwts command line tool can be used to create and manage app-specific local JWTs.

using System.Threading.RateLimiting;
using Microsoft.AspNetCore.Authentication;
using Microsoft.Extensions.Primitives;
using WebRateLimitAuth.Models;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddAuthorization();
builder.Services.AddAuthentication("Bearer").AddJwtBearer();

var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);
var jwtPolicyName = "jwt";

builder.Services.AddRateLimiter(limiterOptions =>
{
    limiterOptions.RejectionStatusCode = StatusCodes.Status429TooManyRequests;
    limiterOptions.AddPolicy(policyName: jwtPolicyName, partitioner: httpContext =>
    {
        var accessToken = httpContext.Features.Get<IAuthenticateResultFeature>()?
                              .AuthenticateResult?.Properties?.GetTokenValue("access_token")?.ToString()
                          ?? string.Empty;

        if (!StringValues.IsNullOrEmpty(accessToken))
        {
            return RateLimitPartition.GetTokenBucketLimiter(accessToken, _ =>
                new TokenBucketRateLimiterOptions
                {
                    TokenLimit = myOptions.TokenLimit2,
                    QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                    QueueLimit = myOptions.QueueLimit,
                    ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod),
                    TokensPerPeriod = myOptions.TokensPerPeriod,
                    AutoReplenishment = myOptions.AutoReplenishment
                });
        }

        return RateLimitPartition.GetTokenBucketLimiter("Anon", _ =>
            new TokenBucketRateLimiterOptions
            {
                TokenLimit = myOptions.TokenLimit,
                QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                QueueLimit = myOptions.QueueLimit,
                ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod),
                TokensPerPeriod = myOptions.TokensPerPeriod,
                AutoReplenishment = true
            });
    });
});

var app = builder.Build();

app.UseAuthorization();
app.UseRateLimiter();

app.MapGet("/", () => "Hello, World!");

app.MapGet("/jwt", (HttpContext context) => $"Hello {GetUserEndPointMethod(context)}")
    .RequireRateLimiting(jwtPolicyName)
    .RequireAuthorization();

app.MapPost("/post", (HttpContext context) => $"Hello {GetUserEndPointMethod(context)}")
    .RequireRateLimiting(jwtPolicyName)
    .RequireAuthorization();

app.Run();

static string GetUserEndPointMethod(HttpContext context) =>
    $"Hello {context.User.Identity?.Name ?? "Anonymous"} " +
    $"Endpoint:{context.Request.Path} Method: {context.Request.Method}";

Limiter with ConcurrencyLimiter, TokenBucketRateLimiter, and authorization

The following sample:

var getPolicyName = "get";
var postPolicyName = "post";
var myOptions = new MyRateLimitOptions();
builder.Configuration.GetSection(MyRateLimitOptions.MyRateLimit).Bind(myOptions);

builder.Services.AddRateLimiter(_ => _
    .AddConcurrencyLimiter(policyName: getPolicyName, options =>
    {
        options.PermitLimit = myOptions.PermitLimit;
        options.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        options.QueueLimit = myOptions.QueueLimit;
    })
    .AddPolicy(policyName: postPolicyName, partitioner: httpContext =>
    {
        string userName = httpContext.User.Identity?.Name ?? string.Empty;

        if (!StringValues.IsNullOrEmpty(userName))
        {
            return RateLimitPartition.GetTokenBucketLimiter(userName, _ =>
                new TokenBucketRateLimiterOptions
                {
                    TokenLimit = myOptions.TokenLimit2,
                    QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                    QueueLimit = myOptions.QueueLimit,
                    ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod),
                    TokensPerPeriod = myOptions.TokensPerPeriod,
                    AutoReplenishment = myOptions.AutoReplenishment
                });
        }

        return RateLimitPartition.GetTokenBucketLimiter("Anon", _ =>
            new TokenBucketRateLimiterOptions
            {
                TokenLimit = myOptions.TokenLimit,
                QueueProcessingOrder = QueueProcessingOrder.OldestFirst,
                QueueLimit = myOptions.QueueLimit,
                ReplenishmentPeriod = TimeSpan.FromSeconds(myOptions.ReplenishmentPeriod),
                TokensPerPeriod = myOptions.TokensPerPeriod,
                AutoReplenishment = true
            });
    }));

See the samples repository for the complete Program.cs file.

Testing endpoints with rate limiting

Before deploying an app using rate limiting to production, stress test the app to validate the rate limiters and options used. For example, create a JMeter script with a tool like BlazeMeter or Apache JMeter HTTP(S) Test Script Recorder and load the script to Azure Load Testing.

Creating partitions with user input makes the app vulnerable to Denial of Service (DoS) Attacks. For example, creating partitions on client IP addresses makes the app vulnerable to Denial of Service Attacks that employ IP Source Address Spoofing. For more information, see BCP 38 RFC 2827 Network Ingress Filtering: Defeating Denial of Service Attacks that employ IP Source Address Spoofing.

Additional resources