.Net Tips – Using custom ServiceThrottlingAttribute to specify WCF service throttling behaviour

If you have created a WCF service in the past then I assume you’re aware that WCF is very heavily configuration-driven and that you can specify the service behaviour including the throttling parameters (MaxConcurrentCalls, MaxConcurrentInstances, MaxConcurrentSessions) in the config file.
But to specify the type of service (PerCall, PerSession or Singleton) you need to apply the ServiceBehaviour attribute to your type.
Now, wouldn’t it be nice if you can specify the service throttling behaviour as an attribute as well rather than having to rely on config files?
I can see the reasoning behind putting these in the config file so you don’t have to recompile and redistribute whenever you wishes to change the throttling behaviour, but there are certain edge cases where it warrants the use of such attribute. For instance, my WCF services are dependency injected through a custom wrapper and I want different throttling behaviour for each service rather than a carpet throttling behaviour which applies to all my services.

ServiceThrottlingAttribute

Whilst there’s no built-in ServiceThrottlingAttribute in the framework, you can easily write one yourself. Here’s one I wrote the other day based on Ralph Squillace’s code sample here:
I made some modification so you don’t HAVE TO specify the value for each type of throttling and will use the default values where possible.
[AttributeUsage(AttributeTargets.Class, AllowMultiple = false)]
public sealed class ServiceThrottlingAttribute : Attribute, IServiceBehavior
{
    public int MaxConcurrentCalls { get; set; }
    public int MaxConcurrentInstances { get; set; }
    public int MaxConcurrentSessions { get; set; }
    public void Validate(ServiceDescription serviceDescription, ServiceHostBase serviceHostBase)
    {
    }
    public void AddBindingParameters(ServiceDescription serviceDescription, ServiceHostBase serviceHostBase, Collection<ServiceEndpoint> endpoints, BindingParameterCollection bindingParameters)
    {
    }
    public void ApplyDispatchBehavior(ServiceDescription serviceDescription, ServiceHostBase serviceHostBase)
    {
        var currentThrottle = serviceDescription.Behaviors.Find<ServiceThrottlingBehavior>();
        if (currentThrottle == null)
        {
            serviceDescription.Behaviors.Add(GetConfiguredServiceThrottlingBehaviour());
        }
    }

    private ServiceThrottlingBehavior GetConfiguredServiceThrottlingBehaviour()
    {
        var behaviour = new ServiceThrottlingBehavior();
        if (MaxConcurrentCalls > 0)
        {
            behaviour.MaxConcurrentCalls = MaxConcurrentCalls;
        }
        if (MaxConcurrentInstances > 0)
        {
            behaviour.MaxConcurrentInstances = MaxConcurrentInstances;
        }
        if (MaxConcurrentSessions > 0)
        {
            behaviour.MaxConcurrentSessions = MaxConcurrentSessions;
        }

        return behaviour;
    }
}

Performance

Before you go ahead and create yourself an ultra-scalable service by setting the max concurrent calls/sessions/instances values to be very high you need to be aware what the performance implications are.

First, you need to know that WCF is configured to be safe from DOS attacks out of the box and the default configurations for service throttling behaviour is very conservative:

  1. MaxConcurrentCalls – default is 16, represents the max number of messages that can actively be processed.
  2. MaxConcurrentInstances – default is 26. For a “PerSession” service this represents the max number of session; for a “PerCall” service this represents the max number of concurrent calls; for a “Singleton” service this is meaningless.
  3. MaxConcurrentSessions – default is 10, represents the max number of sessions a service can accept at one time and only affects session-enabled channels. Increase this to allow more than 10 concurrent sessions.
Prevention of DOS attacks aside, each concurrent request your service is processing also requires at least a new thread to carry out the task. Whilst I don’t know whether WCF uses the framework ThreadPool (which also has further performance implications, see here), there is a balance between the number of concurrent threads and how much time the CPU spends on context switching between the different threads. Too many concurrent threads and the performance starts to suffer as well as the response times, too few concurrent threads and the CPU is under utilised, the response times suffer and more requests are timed out as they wait in the queue.
According to Dan Rigsby’s blog post on throttling WCF services (see references section below), the recommendations for each type of service is as follows:

 

If your InstanceContext is set to “PerCall” you should set maxConcurrentSessions and maxConcurrentCalls to the same value since each call is its own session.  You should target this value to be at least 25-30.  However you shouldn’t need to go higher than 64.

If your InstanceContext is set to “PerSession” you should set maxConcurrentCalls to be at least 25-30.  Your maxConcurrentSessions should be the number of users you want to have concurrently connected.

If your InstanceContext is set to “Single” you should use the same settings as “PerSession”, but the maxConcurrentSessions will only apply if session support is active (this is set with the SessionMode attribute on the ServiceContractAttribute).

References

Dan Rigsby’s blog post on thorttling WCF services