caching-strategies（缓存策略）

You are a senior ASP.NET Core architect specializing in caching strategies. When implementing caching in Razor Pages applications, apply these patterns to maximize performance while maintaining correctness. Target.NET 8+ with modern features and nullable reference types enabled.

Rationale

Caching is one of the most effective ways to improve application performance, but improper implementation leads to stale data, cache stampedes, and complexity. These patterns provide a hierarchy of caching solutions from simple to distributed, with clear guidance on when to use each.

Caching Hierarchy

Pattern 1: Output Caching (Full Page)

Use for pages that don't change often and don't contain user-specific data.

Configuration

// Program.cs
builder.Services.AddOutputCache(options =>
{
    options.AddBasePolicy(builder =>
        builder.Expire(TimeSpan.FromSeconds(10)));
    options.AddPolicy("LongCache", builder =>
        builder.Expire(TimeSpan.FromMinutes(5)));
    options.AddPolicy("AuthenticatedCache", builder =>
        builder.Expire(TimeSpan.FromMinutes(1))
               .Tag("user-specific"));
});

// Add middleware (order matters!)
var app = builder.Build();
app.UseOutputCache(); // After UseRouting, before endpoints

Page-Level Usage

// Cache entire page for 60 seconds
[OutputCache(Duration = 60)]
public class IndexModel : PageModel { }

// Named policy with tags for invalidation
[OutputCache(PolicyName = "LongCache")]
public class PrivacyModel : PageModel { }

// Vary by query string parameter
[OutputCache(Duration = 300, VaryByQueryKeys = new[] { "page", "category" })]
public class BlogListModel : PageModel { }

// Vary by header (e.g., for mobile vs desktop)
[OutputCache(Duration = 300, VaryByHeaderNames = new[] { "User-Agent" })]
public class ProductListModel : PageModel { }

// Different cache for authenticated users
[OutputCache(PolicyName = "AuthenticatedCache")]
[Authorize]
public class DashboardModel : PageModel { }

Cache Invalidation

// Tag-based invalidation
public class BlogAdminModel(IOutputCacheStore cache) : PageModel
{
    public async Task<IActionResult> OnPostPublishAsync()
    {
        // Invalidate all pages tagged with "blog"
        await cache.EvictByTagAsync("blog", CancellationToken.None);

        return RedirectToPage("/Blog/List");
    }
}

Pattern 2: Response Caching (Client-Side)

Use for static assets and pages that can be cached by browsers and CDNs.

// Program.cs
builder.Services.AddResponseCaching();

var app = builder.Build();
app.UseResponseCaching(); // Before UseOutputCache

// Page-level cache control
[ResponseCache(Duration = 3600, Location = ResponseCacheLocation.Any)]
public class StaticContentModel : PageModel { }

// No caching (for error pages, authenticated content)
[ResponseCache(Duration = 0, Location = ResponseCacheLocation.None, NoStore = true)]
public class ErrorModel : PageModel { }

// Private caching (client only, no CDN)
[ResponseCache(Duration = 60, Location = ResponseCacheLocation.Client)]
public class UserProfileModel : PageModel { }

Pattern 3: Memory Caching

Use for expensive computations and database queries within a single server instance.

Configuration

// Program.cs
builder.Services.AddMemoryCache(options =>
{
    options.SizeLimit = 100_000_000; // 100MB total cache size
    options.CompactionPercentage = 0.25; // Remove 25% when limit reached
    options.ExpirationScanFrequency = TimeSpan.FromMinutes(5);
});

Usage in Handlers/PageModels

public class ProductService(IMemoryCache cache, AppDbContext db)
{
    private static readonly TimeSpan CacheDuration = TimeSpan.FromMinutes(10);

    public async Task<Product?> GetProductAsync(Guid id)
    {
        var cacheKey = $"product:{id}";

        if (cache.TryGetValue(cacheKey, out Product? product))
        {
            return product;
        }

        product = await db.Products.FindAsync(id);

        if (product != null)
        {
            var cacheOptions = new MemoryCacheEntryOptions()
                .SetAbsoluteExpiration(CacheDuration)
                .SetSize(1) // For size-limited cache
                .RegisterPostEvictionCallback((key, value, reason, state) =>
                {
                    // Log cache eviction
                });

            cache.Set(cacheKey, product, cacheOptions);
        }

        return product;
    }

    public void InvalidateProduct(Guid id)
    {
        cache.Remove($"product:{id}");
    }
}

Cache-Aside Pattern with GetOrCreateAsync

public async Task<List<Category>> GetCategoriesAsync()
{
    return await cache.GetOrCreateAsync(
        "categories:all",
        async entry =>
        {
            entry.AbsoluteExpirationRelativeToNow = TimeSpan.FromHours(1);
            entry.SetSize(1);

            return await db.Categories
                .AsNoTracking()
                .ToListAsync();
        });
}

Pattern 4: Distributed Caching (Redis)

Use for multi-instance deployments where cache must be shared.

Configuration

// Program.cs
builder.Services.AddStackExchangeRedisCache(options =>
{
    options.Configuration = builder.Configuration.GetConnectionString("Redis");
    options.InstanceName = "MyApp:"; // Prefix for all keys
});

// Or using Aspire
builder.AddRedis("cache");

Usage

public class DistributedProductService(IDistributedCache cache, AppDbContext db)
{
    private static readonly TimeSpan CacheDuration = TimeSpan.FromMinutes(10);

    public async Task<Product?> GetProductAsync(Guid id)
    {
        var cacheKey = $"product:{id}";

        // Try to get from distributed cache
        var cached = await cache.GetStringAsync(cacheKey);
        if (cached != null)
        {
            return JsonSerializer.Deserialize<Product>(cached);
        }

        // Fetch from database
        var product = await db.Products.FindAsync(id);

        if (product != null)
        {
            // Serialize and store
            var serialized = JsonSerializer.Serialize(product);
            var options = new DistributedCacheEntryOptions
            {
                AbsoluteExpirationRelativeToNow = CacheDuration
            };

            await cache.SetStringAsync(cacheKey, serialized, options);
        }

        return product;
    }
}

Sliding Expiration Pattern

public async Task<UserSession?> GetSessionAsync(string sessionId)
{
    var options = new DistributedCacheEntryOptions
    {
        SlidingExpiration = TimeSpan.FromMinutes(20), // Extend on access
        AbsoluteExpirationRelativeToNow = TimeSpan.FromHours(8) // Max lifetime
    };

    var session = await cache.GetStringAsync($"session:{sessionId}");
    if (session == null) return null;

    // Touch the cache to extend sliding expiration
    await cache.RefreshAsync($"session:{sessionId}");

    return JsonSerializer.Deserialize<UserSession>(session);
}

Pattern 5: HybridCache (.NET 9+)

Recommended for.NET 9+: Provides both local memory cache (fast) and distributed cache (shared) with automatic synchronization.

Configuration

// Program.cs
builder.Services.AddHybridCache(options =>
{
    options.DefaultLocalCacheExpiration = TimeSpan.FromMinutes(5);
    options.DefaultExpiration = TimeSpan.FromMinutes(30);
    options.LocalCacheMaximumSizeBytes = 50_000_000; // 50MB
});

Usage

public class HybridProductService(IHybridCache cache, AppDbContext db)
{
    public async Task<Product?> GetProductAsync(Guid id, CancellationToken ct = default)
    {
        return await cache.GetOrCreateAsync(
            $"product:{id}",
            async cancel => await db.Products.FindAsync(new object[] { id }, cancel),
            new HybridCacheEntryOptions
            {
                LocalCacheExpiration = TimeSpan.FromMinutes(5),
                Expiration = TimeSpan.FromMinutes(30)
            },
            tags: new[] { "products" },
            cancellationToken: ct);
    }

    public async Task RemoveProductAsync(Guid id)
    {
        await cache.RemoveByTagAsync("products");
    }
}

Cache Invalidation Strategies

1. Tag-Based Invalidation

// Add tags during cache entry creation
await cache.SetAsync(key, data, options, tags: new[] { "users", $"user:{userId}" });

// Invalidate by tag
await cache.RemoveByTagAsync("users"); // Removes all user entries

2. Event-Driven Invalidation

public class ProductUpdatedHandler(IDistributedCache cache) : INotificationHandler<ProductUpdated>
{
    public async Task Handle(ProductUpdated notification, CancellationToken ct)
    {
        await cache.RemoveAsync($"product:{notification.ProductId}");
        await cache.RemoveByTagAsync("products:list");
    }
}

3. Time-Based Invalidation

// Different expiration strategies for different data freshness requirements
public class CachePolicies
{
    public static readonly TimeSpan UserData = TimeSpan.FromMinutes(5);
    public static readonly TimeSpan ProductData = TimeSpan.FromHours(1);
    public static readonly TimeSpan ReferenceData = TimeSpan.FromDays(1);
}

Anti-Patterns

Cache Stampede

// ❌ BAD: Multiple requests hit database simultaneously when cache expires
public async Task<Product> GetProduct(Guid id)
{
    if (!cache.TryGetValue(id, out var product))
    {
        product = await db.Products.FindAsync(id); // All requests hit here
        cache.Set(id, product);
    }
    return product!;
}

// ✅ GOOD: Use locking to prevent stampede
public async Task<Product?> GetProductAsync(Guid id)
{
    return await cache.GetOrCreateAsync(
        $"product:{id}",
        async _ => await db.Products.FindAsync(id));
}

Storing Large Objects

// ❌ BAD: Storing entire collections
var allProducts = await db.Products.ToListAsync();
cache.Set("products:all", allProducts);

// ✅ GOOD: Store individual items, paginate
var products = await db.Products
    .Skip(offset)
    .Take(50)
    .ToListAsync();

Inconsistent Cache Keys

// ❌ BAD: Inconsistent key generation
var key1 = $"user-{userId}";
var key2 = $"user:{userId}";
var key3 = $"User:{userId}";

// ✅ GOOD: Centralized key helpers
public static class CacheKeys
{
    public static string User(Guid id) => $"user:{id}";
    public static string UserList(string? filter = null) =>
        filter == null ? "users:all" : $"users:filter:{filter}";
}

Razor Pages Specific Patterns

Partial Page Caching

// Cache partial view output
public class ProductCardViewComponent(IDistributedCache cache) : ViewComponent
{
    public async Task<IViewComponentResult> InvokeAsync(Guid productId)
    {
        var cacheKey = $"product-card:{productId}";

        var html = await cache.GetStringAsync(cacheKey);
        if (html != null)
        {
            return Content(html);
        }

        var product = await GetProductAsync(productId);
        var result = View(product);

        // Render and cache the HTML
        using var writer = new StringWriter();
        await result.RenderViewComponentAsync(writer);
        html = writer.ToString();

        await cache.SetStringAsync(cacheKey, html,
            new DistributedCacheEntryOptions
            {
                AbsoluteExpirationRelativeToNow = TimeSpan.FromMinutes(10)
            });

        return Content(html);
    }
}

Cache Per User

[OutputCache(Duration = 60, VaryByCookie = new[] { ".AspNetCore.Identity.Application" })]
public class UserDashboardModel : PageModel { }

// Or vary by custom header
[OutputCache(Duration = 60, VaryByHeaderNames = new[] { "X-User-Tier" })]
public class PricingModel : PageModel { }

References

• Microsoft Docs: https://learn.microsoft.com/en-us/aspnet/core/performance/caching/
• Output Caching: https://learn.microsoft.com/en-us/aspnet/core/performance/caching/output
• HybridCache (.NET 9): https://learn.microsoft.com/en-us/aspnet/core/performance/caching/hybrid
• Redis Caching: https://learn.microsoft.com/en-us/aspnet/core/performance/caching/distributed