How to add an in-memory and a Redis-powered cache layer with ASP.NET Core
29 Jun 2020 #tutorial #asp.net #csharpLet’s say we have a SlowService
that calls a microservice and we need to speed it up. Let’s see how to add a caching layer to a service using ASP.NET Core 6.0.
A cache is a storage layer used to speed up future requests. Reading from a cache is faster than computing data or retrieving it from an external source on every request. ASP.NET Core has built-in abstractions for a caching layer using memory and Redis.
1. In-Memory cache
Let’s start with an ASP.NET Core 6.0 API project with a controller that uses our SlowService
class.
First, let’s install the Microsoft.Extensions.Caching.Memory
NuGet package. Then, let’s register the in-memory cache using the AddMemoryCache()
method.
In our Program.cs
file, let’s do this,
var builder = WebApplication.CreateBuilder(args);
builder.Services.AddControllers();
builder.Services.AddTransient<ISlowService, SlowService>();
builder.Services.AddMemoryCache(options =>
// ^^^^^
{
options.SizeLimit = 1_024;
// ^^^^^
});
var app = builder.Build();
app.MapControllers();
app.Run();
Since memory isn’t infinite, we need to limit the number of items stored in the cache. Let’s use SizeLimit
. It sets the number of “slots” or “places” the cache can hold. Also, we need to tell how many “places” a cache entry takes when stored. More on that later!
Decorate a service to add caching
Next, let’s use the decorator pattern to add caching to the existing SlowService
without modifying it.
To do that, let’s create a new CachedSlowService
. It should inherit from the same interface as SlowService
. That’s the trick!
The CachedSlowService
needs a constructor receiving IMemoryCache
and ISlowService
. This last parameter will hold a reference to the existing SlowService
.
Then, inside the decorator, we will call the existing service if we don’t have a cached value.
public class CachedSlowService : ISlowService
{
private readonly IMemoryCache _cache;
private readonly ISlowService _slowService;
public CachedSlowService(IMemoryCache cache, ISlowService SlowService)
// ^^^^^
{
_cache = cache;
_slowService = slowService;
}
public async Task<Something> DoSomethingSlowlyAsync(int someId)
{
var key = $"{nameof(someId)}:{someId}";
return await _cache.GetOrSetValueAsync(
// ^^^^^
key,
() => _slowService.DoSomethingSlowlyAsync(someId));
}
}
Set Size, Limits, and Expiration Time
Let’s always use expiration times when caching items.
Let’s choose between sliding and absolute expiration times:
SlidingExpiration
resets the expiration time every time an entry is used before it expires.AbsoluteExpirationRelativeToNow
expires an entry after a fixed time, no matter how many times it’s been used.- If we use both, the entry expires when the first of the two times expire
Let’s always add a size to each cache entry. This Size
tells how many “places” from SizeLimit
an entry takes.
When the SizeLimit
value is reached, the cache won’t store new entries until some expire.
Now that we know about expiring entries, let’s create the GetOrSetValueAsync()
extension method. It checks first if a key is in the cache. Otherwise, it uses a factory method to compute and store a value into the cache. This method receives a custom MemoryCacheEntryOptions
to overwrite the default values.
public static class MemoryCacheExtensions
{
// Make sure to adjust these values to suit your own defaults...
public static readonly MemoryCacheEntryOptions DefaultMemoryCacheEntryOptions
= new MemoryCacheEntryOptions
{
AbsoluteExpirationRelativeToNow = TimeSpan.FromSeconds(60),
// ^^^^^
SlidingExpiration = TimeSpan.FromSeconds(10),
// ^^^^^
Size = 1
// ^^^^^
};
public static async Task<TObject> GetOrSetValueAsync<TObject>(
this IMemoryCache cache,
string key,
Func<Task<TObject>> factory,
MemoryCacheEntryOptions options = null)
where TObject : class
{
if (cache.TryGetValue(key, out object value))
{
return value as TObject;
}
var result = await factory();
options ??= DefaultMemoryCacheEntryOptions;
cache.Set(key, result, options);
return result;
}
}
Register a decorated service
To start using the new CachedSlowService
, let’s register it into the dependency container.
Let’s register the existing SlowService
and the new decorated service,
var builder = WebApplication.CreateBuilder(args);
builder.Services.AddControllers();
// Before:
//builder.Services.AddTransient<ISlowService, SlowService>();
// After:
builder.Services.AddTransient<SlowService>();
// ^^^^^
builder.Services.AddTransient<ISlowService>(provider =>
// ^^^^^
{
var cache = provider.GetRequiredService<IMemoryCache>();
var slowService = provider.GetRequiredService<SlowService>();
return new CachedSlowService(cache, SlowService);
// ^^^^^
});
builder.Services.AddMemoryCache(options =>
{
options.SizeLimit = 1_024;
});
var app = builder.Build();
app.MapControllers();
app.Run();
As an alternative, we can use Scrutor, an “assembly scanning and decoration” library, to register our decorators.
Let’s use the Remove()
method to delete cached entries if needed. We don’t want to use outdated or deleted values read from our cache by mistake.
There are only two hard things in Computer Science: cache invalidation and naming things.
– Phil Karlton
From TwoHardThings
Unit Test a decorated service
Let’s see how to test our decorator.
We need a fake for our decorator and assert it’s called only once after two consecutive calls. Let’s use Moq to create fakes.
[TestClass]
public class CachedSlowServiceTests
{
[TestMethod]
public async Task DoSomethingSlowlyAsync_ByDefault_UsesCachedValues()
{
var cacheOptions = Options.Create(new MemoryCacheOptions());
var memoryCache = new MemoryCache(cacheOptions);
// ^^^^^
var fakeSlowService = new Mock<ISlowService>();
fakeSlowService
.Setup(t => t.DoSomethingSlowlyAsync(It.IsAny<int>()))
.ReturnsAsync(new Something());
var service = new CachedSlowService(memoryCache, fakeSlowService.Object);
// ^^^^^
var someId = 1;
await service.DoSomethingSlowlyAsync(someId);
await service.DoSomethingSlowlyAsync(someId);
// ^^^^^
// Yeap! Twice!
fakeSlowService.Verify(t => t.DoSomethingSlowlyAsync(someId), Times.Once);
// Yeap! Times.Once!
}
}
Now, let’s move to the distribute cache.
2. Distributed cache with Redis
A distributed cache layer lives in a separate server. We aren’t limited to the memory of our application server.
A distributed cache is helpful when we share our cache server among many applications or our application runs behind a load balancer.
Redis and ASP.NET Core
Redis is “an open source (BSD licensed), in-memory data structure store, used as a database, cache, and message broker.” ASP.NET Core supports distributed caching with Redis.
Using a distributed cache with Redis is like using the in-memory implementation. We need the Microsoft.Extensions.Caching.StackExchangeRedis
NuGet package and the AddStackExchangeRedisCache()
method.
Now our CachedSlowService
should depend on IDistributedCache
instead of IMemoryCache
.
Also we need a Redis connection string and an optional InstanceName
. With an InstanceName
, we group cache entries with a prefix.
Let’s register a distributed cache with Redis like this,
var builder = WebApplication.CreateBuilder(args);
builder.Services.AddControllers();
builder.Services.AddTransient<SlowService>();
builder.Services.AddTransient<ISlowService>(provider =>
{
var cache = provider.GetRequiredService<IDistributedCache>();
// ^^^^^
var slowService = provider.GetRequiredService<SlowService>();
return new CachedSlowService(cache, SlowService);
// ^^^^^
});
builder.Services.AddStackExchangeRedisCache(options =>
// ^^^^^
{
options.Configuration = "localhost";
// ^^^^^
// I know, I know! We should put it in an appsettings.json
// file instead.
var assemblyName = Assembly.GetExecutingAssembly().GetName();
options.InstanceName = assemblyName.Name;
// ^^^^^
});
var app = builder.Build();
app.MapControllers();
app.Run();
It’s a good idea to read our Redis connection string from a configuration file instead of hardcoding one.
In previous versions of ASP.NET Core, we also had the Microsoft.Extensions.Caching.Redis
NuGet package. It’s deprecated. It uses an older version of the StackExchange.Redis client.
Redecorate a service
Let’s change our CachedSlowService
to use IDistributedCache
instead of IMemoryCache
,
public class CachedSlowService : ISlowService
{
private readonly IDistributedCache _cache;
private readonly ISlowService _slowService;
public CachedSlowService(IDistributedCache cache, ISlowService slowService)
// ^^^^^
{
_cache = cache;
_slowService = slowService;
}
public async Task<Something> DoSomethingSlowlyAsync(int someId)
{
var key = $"{nameof(someId)}:{someId}";
return await _cache.GetOrSetValueAsync(
key,
() => _slowService.DoSomethingSlowlyAsync(someId));
}
}
Now let’s create a new GetOrSetValueAsync()
extension method to use IDistributedCache
instead.
This time, we need the GetStringAsync()
and SetStringAsync()
methods. Also, we need a serializer to cache objects. Let’s use Newtonsoft.Json.
public static class DistributedCacheExtensions
{
public static readonly DistributedCacheEntryOptions DefaultDistributedCacheEntryOptions
= new DistributedCacheEntryOptions
{
AbsoluteExpirationRelativeToNow = TimeSpan.FromSeconds(60),
// ^^^^^
SlidingExpiration = TimeSpan.FromSeconds(10),
// ^^^^^
// We don't need Size here anymore...
};
public static async Task<TObject> GetOrSetValueAsync<TObject>(
this IDistributedCache cache,
string key,
Func<Task<TObject>> factory,
DistributedCacheEntryOptions options = null)
where TObject : class
{
var result = await cache.GetValueAsync<TObject>(key);
if (result != null)
{
return result;
}
result = await factory();
await cache.SetValueAsync(key, result, options);
return result;
}
private static async Task<TObject> GetValueAsync<TObject>(
this IDistributedCache cache,
string key)
where TObject : class
{
var data = await cache.GetStringAsync(key);
if (data == null)
{
return default;
}
return JsonConvert.DeserializeObject<TObject>(data);
}
private static async Task SetValueAsync<TObject>(
this IDistributedCache cache,
string key,
TObject value,
DistributedCacheEntryOptions options = null)
where TObject : class
{
var data = JsonConvert.SerializeObject(value);
await cache.SetStringAsync(key, data, options ?? DefaultDistributedCacheEntryOptions);
}
}
With IDistributedCache
, we don’t need sizes in the DistributedCacheEntryOptions
when caching entries.
Unit Test a decorated service
For unit testing, let’s use MemoryDistributedCache
, an in-memory implementation of IDistributedCache
. This way, we don’t need a Redis server in our unit tests.
Let’s replace the MemoryCache
dependency with the MemoryDistributedCache
like this,
var cacheOptions = Options.Create(new MemoryDistributedCacheOptions());
var memoryCache = new MemoryDistributedCache(cacheOptions);
With this change, our unit test now looks like this,
[TestClass]
public class CachedSlowServiceTests
{
[TestMethod]
public async Task DoSomethingSlowlyAsync_ByDefault_UsesCachedValues()
{
var cacheOptions = Options.Create(new MemoryDistributedCacheOptions());
var memoryCache = new MemoryDistributedCache(cacheOptions);
// ^^^^^
// This time, we're using an in-memory implementation
// of IDistributedCache
var fakeSlowService = new Mock<ISlowService>();
fakeSlowService
.Setup(t => t.DoSomethingSlowlyAsync(It.IsAny<int>()))
.ReturnsAsync(new Something());
var service = new CachedSlowService(memoryCache, fakeSlowService.Object);
// ^^^^^
var someId = 1;
await service.DoSomethingSlowlyAsync(someId);
await service.DoSomethingSlowlyAsync(someId);
// Yeap! Twice again!
fakeSlowService.Verify(t => t.DoSomethingSlowlyAsync(someId), Times.Once);
}
}
We don’t need that many changes to migrate from the in-memory to the Redis implementation.
Conclusion
Voilà! That’s how we cache the results of a slow service using an in-memory and a distributed cache with ASP.NET Core 6.0. Additionally, we can turn on or off the caching layer with a toggle in our appsettings.json
file to create a decorated or raw service.
For more ASP.NET Core content, read how to compress responses and how to serialize dictionary keys. To read more about unit testing, check my Unit Testing 101 guide where I share what I’ve learned about unit testing all these years.
Happy caching time!