In this post I'll talk about a very simple cache implementation for DD4T using the standard .NET System.Web.Caching.Cache class and that allows NULL cached values. The cache itself has no invalidation mechanism and it solely uses a Time-To-Live (TTL) mechanism to expire/evict cached elements.
I used the entire cache wrapper that Rob graciously shared in his post Labels In Dynamic Delivery For Tridion:
The simplified code above attempts to cast the cacheElement to the generic type T that is a specification of ModelBase. If the cast fails, a null model is returned. However, when the cast succeeds, we know that a valid model has been cached and now retrieved from cache.
Of course this implies there is a binding in Ninject from ICacheWrapper to the implementing WebCacheWrapper:
I used the entire cache wrapper that Rob graciously shared in his post Labels In Dynamic Delivery For Tridion:
- The ICacheWrapper interface that defines the cache method signatures;
- The WebCacheWrapper class implementing the ICacheWrapper interface;
In its standard form, Rob's WebCacheWrapper does not allow NULL values to be stored in the cache. Or better yet, it allows them to be stored, but when retrieved, the TryGet method will return false indicating a cache miss.
I personally am a big fan of allowing NULLs to be cached. Take for example, the ability to save the result of a costly operation that resulted in NULL, such as a database lookup for some entity that doesn't actually exist in the DB. In a high-availability environment, one must improve the performance of every single call, thus DB checks that return null must be cached as well.
In order to allow NULLs in DB, I resorted to store something else instead in the cache. Something meaningless in size (1 bit if possible) that would indicate my pseudo-NULL. I chose to store a boolean false as an indication of NULL.
When retrieving the cached value, I'm simply trying to cast the value to the type I requested. If the cast fails (due to the fact that boolean false can't be cast to another object), then I assume a NULL was cached.
The sample code below is from the ModelFactory class discussed in an earlier post. The goal is to cache built models, so that we don't have to load them from the Content Delivery database and convert them into strongly-built models on every invocation of the ModelFactory method.
public T GetModel<T>(string
componentUri) where T : ModelBase
{
T model = null;
object cacheElement;
string key = BuildTheKey(componentUri);
if
(CacheWrapper.TryGet(key, out cacheElement))
{
model = cacheElement as T;
}
else
{
// fetch the actual model
if
(model == null)
{
CacheWrapper.Insert(key, false, 1);
}
else
{
CacheWrapper.Insert(key, model,
cacheMinutes);
}
}
return model;
}
Upon inserting a model into the cache, we check whether the model is null and if so, then we insert a boolean false with a TTL of 1 minute. Otherwise, the actual model is cached using the pre-configured TTL cacheMinutes.
The property CacheWrapper is of type ICacheWrapper and it is injected into the ModelFactory class using Ninject. More information about dependency injection in DD4T .net, in this post:
[Inject]
public virtual ICacheWrapper CacheWrapper { get; set; }
Bind<ICacheWrapper>().ToMethod(context =>
new WebCacheWrapper(HttpRuntime.Cache)).InSingletonScope();
Comments