Performance Tuning in ASP.NET Core: Best Practices for 2025

Arulraj Aboorvasamy

14 min read Dec 2, 2025

Summarize this blog post with:

TL;DR: Supercharge ASP.NET Core 10 apps with five key optimizations: cut asset sizes by up to 92% using MapStaticAssets, accelerate caching with HybridCache, enforce rate limiting and timeouts for stability, slim Blazor bundles with AOT and WasmStripILAfterAOT, and track performance in real time with expanded metrics. Ready to boost speed and resilience? Explore the full guide now!

As web applications demand higher speed and scalability, .NET 10 introduces major performance improvements for ASP.NET Core. This guide explains the latest optimizations and practical steps to help you:

Maximize throughput
Reduce latency
Minimize resource usage

Whether you’re building APIs, real-time apps, or Blazor UIs, these techniques will keep your apps fast and efficient.

Why performance tuning matters in 2025

Modern applications face increasing traffic and complex UI demands. Without performance tuning, load times and infrastructure expenses increase. .NET 10 delivers advanced solutions for assets, caching, and rendering challenges.

Agenda

This guide presents five essential techniques that make ASP.NET Core apps faster, lighter, and secure.

Optimize static asset delivery with MapStaticAssets and client-side fingerprinting.
Leverage the HybridCache for efficient caching.
Secure rate limiting and timeouts.
Trim IL after AOT: Reduce bundle size in WASM.
Monitor and profile with enhanced metrics.

Note: If you are new to ASP.NET Core, check the blog ASP.NET Core 3.0 Performance Optimization, where you can find a few tips that are still useful in the latest versions as well.

1. Optimize static asset delivery with MapStaticAssets and client-side fingerprinting

.NET 9 introduced MapStaticAssets, replacing the UseStaticFiles method. It improves static asset delivery with build-time compression, gzip during development, gzip + Brotli in production. and SHA-256 ETags for better caching.

In .NET 10, this optimization is extended to client-side fingerprinting for JavaScript modules in standalone Blazor WASM apps. This new update further enhances performance by ensuring robust browser caching and efficient resource delivery.

Benchmarks: Default Razor pages template size drops from 331.1 KB to 65.5 KB (80.20% reduction). For Blazor apps, Fluent UI shrinks from 478 KB to 84 KB (82.43%), and MudBlazor from 588.4 KB to 46.7 KB (92.07%).
Implementation: To achieve the above metrics, replace the app.UseStaticFiles() method with the app.MapStaticAssets() method in the Program.cs file. This simplifies delivery and minimizes runtime overhead, ideal for static-heavy apps like Blazor web apps. Refer to the following code example.
```
// app.UseStaticFiles(); 
app.MapStaticAssets();
```

To enable pre-compression and fingerprinting for all static assets, we need to reduce runtime overhead.

If you are using Blazor WebAssembly, enable client-side fingerprinting. To do so, update the wwwroot/index.html file to add the fingerprint placeholders.

<head>
    <script type="importmap"></script>
</head>
<body>
    <script src="_framework/blazor.webassembly#[.{fingerprint}].js"></script>
</body>

In the project file (.csproj), add the following code:

<PropertyGroup>
    <OverrideHtmlAssetPlaceholders>true</OverrideHtmlAssetPlaceholders>
</PropertyGroup>

For additional JavaScript modules, use the <StaticWebAssetFingerprintPattern> property to fingerprint specific file types (e.g., .mjs).

<StaticWebAssetFingerprintPattern Include="JSModule" Pattern="*.mjs" Expression="#[.{fingerprint}]!" />

This lets the framework automatically integrate fingerprinted files into the import map for browser resolution.

Preloading assets in Blazor web apps

In .NET 10, Blazor web apps use the ResourcePreloader component to preload framework static assets, replacing the <link> headers.

We can add it to the App. Razor file as shown below.

<head>
    <base href="/" />
    <ResourcePreloader />
</head>

This ensures assets are loaded efficiently with correct app base path resolution.

Refer to the following image. Here, the Blazor asset preload fetches critical resources during connection setup, enabling faster rendering.

2. Leverage the HybridCache for efficient caching

The HybridCache library, introduced as a preview in .NET 9 and fully supported after release as Microsoft.Extensions.Caching.Hybrid integrates in-memory and distributed caching with stampede protection to avoid redundant fetches on cache misses. This prevents performance issues in high-concurrency APIs and microservices. It also enables the graceful handling of cancellation tokens to prevent unwanted executions.

Implementation: Use the GetOrCreateAsync method for seamless caching as shown below.

var data = await cache.GetOrCreateAsync("key", async entry => await ExpensiveOperationAsync());

Benefits: Reduces database/API calls, supports custom serializers (default: System.Text.Json), and optimizes immutable types with object reuse, zero garbage collection pressure via the ValueTask<T> method, and pooled buffers.
When to avoid: The HybridCache may not be the best fit for simple, low-traffic apps or real-time systems with strict fresh data requirements. Older projects, before .NET 8, may use the IMemoryCache or IDistributedCache interfaces directly.
Benchmarks: Significant reduction in operation duplication under load.

Best practices

HybridCache is most effective when used for frequently accessed data such as API responses, computed results, or expensive queries. This minimizes backend strain and improves overall responsiveness. The ASP.NET Core 9 and later releases include features such as monitoring cache hit rates with metrics, static asset optimization, and Blazor enhancements that complement HybridCache.

The diagram below illustrates how HybridCache achieves higher effective hit rates by falling back to Redis from the in-memory cache when needed. (Note: Numbers shown are representative, not actual benchmarks.)

Comparing different types of caching solutions

Feature/Aspect	HybridCache	IMemoryCache	IDistributedCache
Package	Microsoft.Extensions.Caching.Hybrid (.NET 8+)	Microsoft.Extensions.Caching.Memory (.NET Core 1.0+)	Microsoft.Extensions.Caching.Abstractions (.NET Core 1.0+)
Caching scope	In-memory by default(L1), with seamless support for distributed caches (L2)(e.g., Redis, SQL Server).	In-memory only; no built-in support for distributed caching.	Distributed caching only (e.g., Redis, SQL Server, NCache); requires external cache store.
API implicity	Simplified API with GetOrCreateAsync for automatic cache population and factory execution.	More manual API; requires explicit Get, Set, and CreateEntry operations.	Low-level API; requires manual GetAsync, SetAsync, and RemoveAsync operations.
Cache stampede protection	Built-in protection to prevent multiple threads from populating the same cache key simultaneously.	No built-in protection; requires manual synchronization (e.g., locks).	No built-in protection; requires manual synchronization or external coordination.
Serialization	Automatic serialization/deserialization of complex objects for both in-memory and distributed caches.	No automatic serialization; objects stored directly in memory, manual handling needed for serialization.	Requires manual serialization/deserialization (e.g., to/from byte arrays).
Cancellation support	Supports CancellationToken in GetOrCreateAsync for canceling expensive operations.	No direct cancellation support; must be handled manually in factory logic.	No direct cancellation support; must be handled manually in factory logic.
Distributed cache integration	Native integration with IDistributedCache providers, seamless fallback to in-memory if no distributed cache is configured.	No direct integration; requires separate use of IDistributedCache with manual coordination.	Core interface for distributed caching; requires specific provider implementation (e.g., Redis, SQL Server).
Performance	Optimized for high-concurrency with stampede protection and efficient distributed cache integration.	Lightweight for simple in-memory scenarios, but less efficient in high-concurrency without manual synchronization.	Depends on the provider (e.g., Redis is fast but adds network latency); no concurrency protections.
Expiration policies	Supports expiration via HybridCacheEntryOptions.	Supports flexible expiration (absolute, sliding, size limits, post-eviction callbacks).	Supports expiration via DistributedCacheEntryOptions, but implementation varies by provider.
Metrics integration	Benefits from ASP.NET Core 10’s metrics (e.g., Microsoft.AspNetCore.MemoryPool) for monitoring cache performance.	Limited metrics integration; requires custom instrumentation.	Limited metrics; depends on provider-specific monitoring (e.g., Redis metrics).
Use cases	Ideal for high-traffic monolithic apps, microservices, or Blazor apps needing robust caching with minimal code.	Best for simple, low-traffic monolithic apps with in-memory caching needs.	Suitable for distributed environments or multi-instance apps requiring shared caching.
Complexity	Higher-level abstraction, reducing boilerplate but requiring configuration for distributed caches.	Lower-level abstraction, offering more control but requiring more code for complex scenarios.	Lowest-level abstraction; requires manual serialization, provider setup, and error handling.
Scalability	Scales seamlessly from single instance to multi-instance with distributed cache support.	Limited to single-instance apps; not scalable to distributed environments without manual integration.	Designed for distributed environments, scalable across multiple instances.

3. Secure rate limiting and timeouts

Securing your apps through rate limiting and timeouts is a cornerstone of performance tuning. They prevent resource exhaustion from malicious traffic or slow operations while maintaining high throughput for legitimate requests.

Rate limiting: It is powered by the built-in Microsoft.AspNetCore.RateLimiting middleware. It enforces policies such as fixed-window limits, e.g., 100 requests per minute or concurrency limits to throttle excessive calls, thereby directly reducing CPU and memory spikes during Distributed Denial of Service (DDoS) scenarios. Hence, it improves performance and reduces service costs.
Request timeouts: It is a middleware from Microsoft.AspNetCore.Http.Timeouts that cap endpoint execution time, such as 30 seconds globally or per-route, to avoid long-running tasks hogging threads, ensuring the Kestrel server remains responsive.

A combination of these two features optimizes overall performance by freeing up server resources for concurrent workloads, thereby enhancing overall efficiency. They also align with .NET 10’s enhanced JIT compilation and runtime efficiencies for faster cold starts and lower latency.

Refer to the following image, which illustrates ASP.NET Core rate-limiting and timeout protection.

Here’s an example implementation in a minimal API setup for Program.cs file.

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

var builder = WebApplication.CreateBuilder(args);

// Configure rate limiting
builder.Services.AddRateLimiter(options =>
{
    options.AddFixedWindowLimiter("fixed", opt =>
    {
        opt.PermitLimit = 100;
        opt.Window = TimeSpan.FromMinutes(1);
        opt.QueueProcessingOrder = QueueProcessingOrder.OldestFirst;
        opt.QueueLimit = 10;
    });
    options.OnRejected = (context, token) =>
    {
        context.HttpContext.Response.StatusCode = 429;
        context.HttpContext.Response.WriteAsync("Too many requests. Try again later.");
        return ValueTask.CompletedTask;
    };
});

// Configure timeouts
builder.Services.AddRequestTimeouts();

var app = builder.Build();

// Apply rate limiting globally
app.UseRateLimiter();

// Apply global timeout
app.UseRequestTimeouts();

// Example endpoint with specific timeout and rate limit
app.MapGet("/api/data", async (HttpContext ctx) =>
{
    await Task.Delay(5000); // Simulate work
    return "Data retrieved";
})
.WithName("DataEndpoint")
.WithRequestTimeout(TimeSpan.FromSeconds(10))
.RequireRateLimiting ("fixed"); // Apply policy, I have set this in options

app.Run();

Pros and cons

Aspect	Pros	Cons
Rate limiting	Enhances security against brute-force attacks and abuse. Improves scalability by distributing load evenly. Low overhead with in-memory partitioning (supports Redis for distributed setups in .NET 10).	Potential for false positives during traffic spikes. Requires load testing to tune limits. In-memory storage isn’t ideal for stateless scaling without extensions.
Timeouts	Prevents thread pool exhaustion from hung requests. Customizable per endpoint for fine-grained control. Integrates seamlessly with async patterns for better responsiveness.	Abrupt cutoffs can frustrate users on slow networks. No built-in retry logic; needs custom handling. Debug mode disables enforcement, risking overlooked issues.

Dos

Load test in a production-like environment with tools like Apache JMeter or Bombardier to benchmark limits. For example, ensure the 99th percentile response time stays under 200ms under burst traffic.
Use partitioners, such as IP addresses or client IDs, for targeted limiting and combine them with authentication for API keys.
Set global timeouts conservatively (e.g., 120 seconds) and override them for long-running endpoints, such as file uploads.

Don’ts

Don’t ignore rejection handling; always return informative 429 responses with Retry-After headers.
Don’t apply overly aggressive limits without monitoring; start with permissive limits and tighten them based on analytics in production or check with a similar system.
Don’t forget to exclude health check endpoints from limits to avoid false alerts in monitoring tools like Application Insights.

4. Trim IL after AOT: Reduce bundle size in WASM

Although the WebAssembly Native AOT (Ahead-of-Time compilation) was introduced in .NET 6, it was experimental. Although it matured from .NET 7 to .NET 9, it was not production-ready. With .NET 10, WebAssembly Native AOT compilation is now production-ready for Blazor apps using Interactive WebAssembly or Auto render modes.

New optimization

To further optimize download size after Native AOT, Microsoft recommends enabling WasmStripILAfterAOT. This MSBuild property removes the .NET Intermediate Language (IL) from compiled methods after AOT compilation.

The result?

Significantly reducing the size of the _framework folder in the WebAssembly app. While most methods can be safely trimmed, some are retained for runtime reflection or interpreter use.

How to enable

Add the following code to your .Client project file.

<PropertyGroup>
    <PublishAot>true</PublishAot>
    <WasmStripILAfterAOT>true</WasmStripILAfterAOT>
</PropertyGroup>

When to use AOT?

Interactive Auto apps
Data-heavy UIs (grids, charts)
Long user sessions
Offline/PWA support

Skip AOT (use JIT)

Simple marketing sites
Frequent deployments
Small bundles critical
Dev/test environments

Refer to the following image. From this, we can conclude that Native AOT has a lower app size, memory usage, and startup time; trimming IL will give more optimization.

Data source: Microsoft

Note: The same flag <PublishAot>true</PublishAot> is also used for Server-side apps. The WasmStripILAfterAOT property is only applicable for WebAssembly in the browser. For compatibility summary, refer to the official documentation.

5. Monitor and profile with enhanced metrics

Observability is the foundation of performance tuning in 2025. ASP.NET Core has evolved from basic logging to a rich, hierarchical metrics ecosystem that lets you measure exactly where time and resources are spent. From the moment your server starts to fine-grained Blazor component rendering, built-in metrics empower you to detect regressions before users do.

Let’s explore the key metrics and tools to master this!

ASP.NET Core 10 identity metrics: Tune authentication at scale

Observability in .NET 10 goes beyond basic logging; it now includes rich metrics for Identity operations, helping you optimize authentication flows and security without sacrificing performance.

What’s new?

ASP.NET Core 10 introduces the Microsoft.AspNetCore.Identity meter, which provides counters, histograms, and gauges to track critical user and session behaviors in real time:

User creation
Password updates
Role assignments
Login attempts
Two-factor authentication usage
Token validation

Key metrics

aspnetcore.identity.sign_in.authenticate.duration: Measures sign-in time
aspnetcore.identity.user.create.duration: Tracks user creation latency
aspnetcore.identity.sign_in.two_factor_clients_remembered: Monitors 2FA usage trends

These metrics enable proactive performance tuning and security monitoring.

Integration

Integrate these metrics with OpenTelemetry or Prometheus to visualize trends, detect anomalies (e.g., spikes in failed logins), and optimize authentication flows, ensuring both speed and safety at scale. For complete setup, refer to the ASP.NET Core metrics documentation.

Here is the table that explains the metrics and what they measure.

Key metrics	What it measures	Why it matters
ServerReady event.	Time from process start to Kestrel listening.	Critical for cold start in containers, serverless.
Kestrel errors, SignalR, Blazor tracing.	Connection drops, hub invocations, circuit lifecycle.	Pinpoint real-time app bottlenecks.
Auth metrics, profiling counters.	Sign-in duration, challenge count, CPU/memory per request.	Secure identity-heavy apps without perf tax.

Tools to profile

Profiling is the compass for performance tuning; it turns guesswork into data-driven wins. ASP.NET Core provides a robust toolchain to help you identify bottlenecks and optimize effectively.

Essential profiling tools

dotnet-counters: Real-time snapshots of GC pauses, CPU spikes, and memory trends during load tests.
dotnet-trace: Captures EventPipe data in production with near-zero overhead. Ideal for diagnosing latency jitter and intermittent issues.
Visual Studio Profiler: Visualizes per-method allocations and hot paths via interactive graphs.
dotnet-dump: Enables post-mortem forensics on crashes without restarting.

Why these tools matter

Together, these tools form a complete observability loop, helping you:

Cut latency by 30%+
Reduce memory usage by 20%
Ship resilient apps at scale

Pro Tip: Profile early, profile often, and let metrics guide every optimization.

Tooling stack for ASP.NET Core 2025

Tool	Use cases
dotnet-counters	Live GC, thread pool, JIT
dotnet-trace	CPU sampling, GC events
Prometheus + Grafana	Long-term dashboards
Aspire Dashboard	.NET 10 native (auth, Blazor, Kestrel)

Note: For more details, refer to the diagnostic tools documentation.

Wrapping up

Thanks for reading! ASP.NET Core performance tuning in 2025 is no longer about isolated tweaks; it’s a holistic discipline. From build-time asset optimization and hybrid caching to runtime rate limiting, AOT trimming, and granular observability, .NET 10 gives you the tools to build apps that are:

80–92% lighter in payload size.
Highly resilient against cache stampedes and traffic spikes.
Blazing fast, with sub-200ms latencies under load.

Whether you’re powering real-time Blazor experiences, high-throughput APIs, or globally scaled microservices, these best practices will help you define modern performance standards.

Next steps:

Apply these optimizations incrementally.
Measure relentlessly with dotnet-counters and OpenTelemetry.
Keep profiling and monitoring at the heart of your workflow.

Your apps won’t just meet expectations; they’ll set the benchmark for speed and scalability.

Start your next-gen ASP.NET Core project today with Syncfusion

Syncfusion offers Day 1 support for .NET 10, ensuring full compatibility and optimized performance from the start. Build and deploy ASP.NET Core apps confidently with Syncfusion’s powerful component suite, and create next-generation applications that are faster, smarter, and more efficient.

Existing Syncfusion users can download the newest version of Essential Studio from the license and download page, while new users can start a 30-day free trial to experience its full potential.

If you have any questions, contact us through our support forum, support portal, or feedback portal. We are always happy to assist you

Tags:

Meet the Author