mcpField
C
caching-strategies

caching-strategies

github.com/wshaddix开发9179 次安装37 周前更新
代码仓库

关于 caching-strategies

面向 ASP.NET Core 架构师的 Razor Pages 缓存实践指南，涵盖缓存层级选择及避免数据过期、缓存击穿和实现复杂化的方法。

主要功能

  1. 01输出缓存与响应缓存模式
  2. 02内存、分布式及 HybridCache 指南
  3. 03缓存失效策略
  4. 04Razor Pages 与按用户缓存
  5. 05反模式及缓存键规范

适用场景

  1. 01为应用选择缓存类型
  2. 02缓存开销较大的数据库查询
  3. 03在多个服务器实例间共享缓存数据
  4. 04更新后清除缓存页面或记录

安装

复制下面的命令，在终端中运行。

$npx skills add https://github.com/wshaddix/dotnet-skills --skill caching-strategies

安装后在 Claude Code 中输入/caching-strategies确认已加载。

详细介绍

你是一位专注于缓存策略的资深 ASP.NET Core 架构师。在 Razor Pages 应用中实现缓存时，请采用这些模式，在保持正确性的同时最大限度地提升性能。目标平台为 .NET 8+，并启用现代功能和可空引用类型。

原理

缓存是提升应用性能最有效的方法之一，但实现不当会导致数据过时、缓存雪崩和复杂性增加。这些模式提供了从简单到分布式的缓存解决方案层级，并明确说明了各自的适用场景。

缓存层级

策略范围适用场景延迟
输出缓存整个服务器完整页面响应低
响应缓存客户端 + 代理静态页面、资源低
内存缓存单个实例生命周期较短的高开销数据非常低
分布式缓存多个实例多台服务器间共享数据低至中
HybridCache（.NET 9+）多个实例兼具内存缓存和分布式缓存的优势非常低

模式 1：输出缓存（完整页面）

适用于不常更改且不包含用户专属数据的页面。

配置
// 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
页面级用法
// 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 { }
缓存失效
// 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");
    }
}

模式 2：响应缓存（客户端）

适用于可由浏览器和 CDN 缓存的静态资源和页面。

// 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 { }

模式 3：内存缓存

适用于单个服务器实例内开销较大的计算和数据库查询。

配置
// 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);
});
在处理程序/PageModel 中使用
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}");
    }
}
使用 GetOrCreateAsync 的 Cache-Aside 模式
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();
        });
}

模式 4：分布式缓存（Redis）

适用于需要跨多个实例共享缓存的部署。

配置
// 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");
用法
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;
    }
}
滑动过期模式
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);
}

模式 5：HybridCache（.NET 9+）

推荐用于 .NET 9+：同时提供本地内存缓存（速度快）和分布式缓存（可共享），并自动进行同步。

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

缓存失效策略

1. 基于标签的失效
// 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. 事件驱动的失效
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. 基于时间的失效
// 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);
}

反模式

缓存雪崩
// ❌ 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));
}
存储大型对象
// ❌ 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();
缓存键不一致
// ❌ 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 专用模式

部分页面缓存
// 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);
    }
}
按用户缓存
[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 { }

参考资料

评论

支持 Markdown