Caching Providers

As caching is an essential technology in the development of high-performance web services, ServiceStack has a number of different caching options available that each share the same common client interface (ICacheClient) for the following cache providers:

  • Memory Cache - Useful for single host web services without needing any infrastructure dependencies.
  • Redis - A fast key-value store with non-volatile persistent storage and support for rich comp-sci data structures.
  • OrmLiteCacheClient - Supports all OrmLite's RDBMS providers for using an existing RDBMS as a distributed cache.
  • Memcached - The original, tried and tested distributed memory caching provider.
  • Aws DynamoDB - Uses Amazon's Dynamo DB backend hosted on Amazon Web Services
  • Azure Table Storage - Uses Azure Table Storage for when your application is hosted on Azure.

Async Cache Clients

All remote Caching Providers also implement the ICacheClientAsync async APIs whilst any other ICacheClient only providers like the local in-memory MemoryCacheClient are still able to use the ICacheClientAsync interface as they'll return an Async Wrapper over the underlying sync APIs.

So even if you're currently only using MemoryCacheClient or your own ICacheClient sync implementation, you can still use the async Caching Provider API now and easily switch to an async caching provider in future without code changes.

The Async Caching Provider APIs are accessible via the CacheAsync property in ServiceStack Service or ServiceStackController classes, e.g:

public async Task<object> Any(MyRequest request)
{
    var item = await CacheAsync.GetAsync<Item>("key");
    //....
}

public class HomeController : ServiceStackController
{
    public async Task<ActionResult> Index()
    {
        var item = await CacheAsync.GetAsync<Item>("key");
    }
}

Whilst outside of ServiceStack you can AppHost.GetCacheClientAsync(), e.g:

var cache = HostContext.AppHost.GetCacheClientAsync();
var item = await cache.GetAsync<Item>("key");

Configure Caching Providers

To configure which cache should be used, the particular client has to be registered in the IoC container against the ICacheClient interface:

Memory cache:

By default ServiceStack registers an MemoryCacheClient by default when no ICacheClient is registered so no registration is necessary.

//container.Register<ICacheClient>(new MemoryCacheClient());

Even if you have an alternative ICacheClient registered you can still access the in memory cache via the LocalCache property in your Services and ServiceStack MVC Controllers or anywhere else via the HostContext.AppHost.GetMemoryCacheClient() singleton as well as [CacheResponse(UseLocalCache=true)] when using the Cache Response Attribute.

Redis

container.Register<IRedisClientsManager>(c => 
    new RedisManagerPool("localhost:6379"));

container.Register(c => c.Resolve<IRedisClientsManager>().GetCacheClient());
NuGet Package: ServiceStack.Redis

OrmLite

//Register OrmLite Db Factory if not already
container.Register<IDbConnectionFactory>(c => 
    new OrmLiteConnectionFactory(connString, SqlServerDialect.Provider)); 

container.RegisterAs<OrmLiteCacheClient, ICacheClient>();

//Create 'CacheEntry' RDBMS table if it doesn't exist already
container.Resolve<ICacheClient>().InitSchema(); 

SQL Server Memory Optimized Cache

SQL Server's Memory Optimized support can be used to improve the performance of OrmLiteCacheClient by configuring it to use the above In Memory Table Schema instead, e.g:

container.Register<ICacheClient>(c => 
    new OrmLiteCacheClient<SqlServerMemoryOptimizedCacheEntry>());
NuGet Package: ServiceStack.Server

Memcached:

container.Register<ICacheClient>(
    new MemcachedClientCache(new[] { "127.0.0.0" }); //Add Memcached hosts
NuGet Package: ServiceStack.Caching.Memcached

AWS DynamoDB:

var awsDb = new AmazonDynamoDBClient(
    AWS_ACCESS_KEY, AWS_SECRET_KEY, RegionEndpoint.USEast1);

container.Register<IPocoDynamo>(new PocoDynamo(awsDb));
container.Register<ICacheClient>(c => new DynamoDbCacheClient(c.Resolve<IPocoDynamo>()));

var cache = container.Resolve<ICacheClient>();
cache.InitSchema();
NuGet Package: ServiceStack.Aws

Azure:

container.Register<ICacheClient>(new AzureTableCacheClient(cacheConnStr));
NuGet Package: ServiceStack.Azure

Multi CacheClient

The MultiCacheClient can be used to utilize a write-through multi-tiered cache client where all "writes" are made to all registered cache providers whilst "reads" are only accessed until a value exists. E.g. you can register a local memory and redis server backed Cache Client with:

container.Register<ICacheClient>(c => new MultiCacheClient(
    new MemoryCacheClient(),
    c.Resolve<IRedisClientsManager>().GetCacheClient()));

Cache a response of a service

To cache a response you simply have to call ToOptimizedResultUsingCache which is an extension method existing in ServiceStack.ServiceHost.

In your service:

public class OrdersService : Service
{
    public object Get(CachedOrders request)
    {
        var cacheKey = "unique_key_for_this_request";
        return base.Request.ToOptimizedResultUsingCache(base.Cache,cacheKey,()=> 
            {
                //Delegate is executed if item doesn't exist in cache 
                //Any response DTO returned here will be cached automatically
            });
    }
}

Tip

There exists a class named UrnId which provides helper methods to create unique keys for an object

ToOptimizedResultUsingCache also has an overload which provides a parameter to set the timespan when the cache should be deleted (marked as expired). If now a client calls the same service method a second time and the cache expired, the provided delegate, which returns the response DTO, will be executed a second time.

var cacheKey = "some_unique_key";
//Cache should be deleted in 1h
var expireInTimespan = new TimeSpan(1, 0, 0);
return base.Request.ToOptimizedResultUsingCache(
    base.Cache, cacheKey, expireInTimespan, ...)

Delete cached responses

If now for example an order gets updated and the order was cached before the update, the webservice will still return the same result, because the cache doesn't know that the order has been updated.

So there are two options:

  • Use time based caching (and expire cache earlier)
  • Cache on validity

INFO

When the cache is based on validity the caches are invalidated manually (e.g. when a user modified his profile, > clear his cache) which means you always get the latest version and you never need to hit the database again to rehydrate the cache if it hasn't changed, which will save resources

So if the order gets updated, you should delete the cache manually:

public class CachedOrdersService : Service
{
    public async Task Put(CachedOrders request)
    {
        //The order gets updated...
        var cacheKey = "some_unique_key_for_order";
        await CacheAsync.ClearCachesAsync(cacheKey);
    }
}

If now the client calls the webservice to request the order, he'll get the latest version.

LocalCache

As it sometimes beneficial to have access to a local in-memory Cache in addition to your registered ICacheClient Caching Provider we also pre-register a MemoryCacheClient that all your Services now have access to from the LocalCache property, i.e:

MemoryCacheClient LocalCache { get; }

This doesn't affect any existing functionality that utilizes a cache like Sessions which continue to use your registered ICacheClient, but it does let you change which cache you want different responses to use, e.g:

var cacheKey = "unique_key_for_this_request";
return base.Request.ToOptimizedResultUsingCache(LocalCache, cacheKey, () => {
    //Delegate is executed if item doesn't exist in cache 
});

Or if you're using the CacheResponse attribute you can specify to cache responses in the local cache with:

[CacheResponse(LocalCache = true)]
public object Any(MyRequest request) { ... }

INFO

If you don't register a ICacheClient ServiceStack automatically registers a MemoryCacheClient for you which will also refer to the same instance registered for LocalCache

ICacheClientExtended

The ICacheClientExtended API is used to to provide additional non-core functionality to our most popular Caching providers:

  • Redis
  • OrmLite RDBMS
  • In Memory
  • AWS
  • Azure

The new API's are added as Extension methods on ICacheClient so they're easily accessible without casting, the new API's available include:

  • GetKeysByPattern(pattern) - return keys matching a wildcard pattern
  • GetAllKeys() - return all keys in the caching provider
  • GetKeysStartingWith() - Streaming API to return all keys Starting with a prefix

With these new API's you can now easily get all active User Sessions using any of the supported Caching providers above with:

var sessionPattern = IdUtils.CreateUrn<IAuthSession>(""); //= urn:iauthsession:
var sessionKeys = Cache.GetKeysStartingWith(sessionPattern).ToList();

var allSessions = Cache.GetAll<IAuthSession>(sessionKeys);

CacheClient with Prefix

The CacheClientWithPrefix class lets you decorate any ICacheClient to prefix all cache keys using the .WithPrefix() extension method. This could be used to easily enable multi-tenant usage of a single redis instance, e.g:

container.Register(c => 
    c.Resolve<IRedisClientsManager>().GetCacheClient().WithPrefix("site1"));

Live Example and code

A live demo of the ICacheClient is available in The ServiceStack.Northwind's example project. Here are some requests to cached services:

Which are simply existing web services wrapped using ICacheClient that are contained in CachedServices.cs