InstanceContextMode Options
InstanceContextMode, set via [ServiceBehavior(InstanceContextMode = InstanceContextMode.X)], controls the lifetime relationship between service instances and client calls. PerCall (the default for wsHttpBinding without sessions) creates a brand-new service instance for every single operation call and disposes it immediately after, which means the service class can never hold state between calls — great for stateless, horizontally scalable services. PerSession creates one service instance per client session and keeps it alive for the duration of that session, letting the service hold conversational state (like a shopping cart accumulated across several calls) in instance fields. Single creates exactly one service instance for the entire application's lifetime, shared by every client and every call, which is useful for scenarios needing a shared in-memory cache or counter, but which makes that single instance a potential bottleneck and a thread-safety hazard if not paired with the right ConcurrencyMode.
Cricket analogy: PerCall hires fresh net bowlers for every single practice ball, PerSession assigns one coach to a batter for the whole tour, and Single is one physiotherapist shared by all 15 squad members all season.
[ServiceBehavior(
InstanceContextMode = InstanceContextMode.PerSession,
ConcurrencyMode = ConcurrencyMode.Single)]
public class ShoppingCartService : IShoppingCartService
{
private readonly List<CartItem> _items = new List<CartItem>();
public void AddItem(CartItem item)
{
// Safe: ConcurrencyMode.Single serializes all access,
// and PerSession means _items persists across calls
// within the same client session.
_items.Add(item);
}
public IEnumerable<CartItem> GetItems() => _items;
}ConcurrencyMode and Thread Safety
ConcurrencyMode, also set on [ServiceBehavior], governs how many threads WCF allows to execute inside a given service instance context simultaneously. Single, the default, serializes all calls into an instance context with a lock — only one thread executes inside that instance at a time, which makes instance fields inherently thread-safe without any manual locking, at the cost of throughput if that instance is called frequently. Multiple allows several threads to execute concurrently inside the same instance, which is necessary for a Single-instanced service to actually scale under load, but it means every mutable field the service touches must be protected with explicit locks, concurrent collections, or Interlocked operations, because the runtime provides zero automatic thread safety. Reentrant, a middle ground, still serializes calls like Single but releases the lock when the operation makes an outbound call (to another service or via a callback) and reacquires it on return, which prevents deadlocks in callback-heavy duplex scenarios but reintroduces the possibility of the instance's state being mutated by another thread while the outbound call is in flight.
Cricket analogy: Single is a catch drill where one fielder takes a ball at a time, Multiple is a multi-ball drill with several coaches throwing at once, and Reentrant is a bowler pausing mid run-up for a rope adjustment, then resuming exactly where he left off.
ConcurrencyMode.Multiple combined with InstanceContextMode.PerCall is often unnecessary complexity: since PerCall already gives every call its own fresh instance with no shared state, there's rarely any mutable instance field to protect, so Multiple's extra concurrency doesn't introduce thread-safety risk in that specific combination — the risk shows up specifically when Multiple is paired with PerSession or Single instancing.
Combining Instancing with Concurrency
The instancing and concurrency modes are configured together, and certain combinations are especially common in practice. InstanceContextMode.Single with ConcurrencyMode.Single is the safest way to build a shared, stateful, in-memory cache service, since every call is fully serialized — but it also means the service's entire throughput is bounded by however long one call takes, so long-running operations on this combination will bottleneck every other client. InstanceContextMode.Single with ConcurrencyMode.Multiple unlocks real concurrency for a shared instance but shifts the thread-safety burden entirely onto the developer, who must treat every field like it's being hit by concurrent requests from unrelated clients simultaneously — a mistake here (an unprotected Dictionary.Add call, for instance) produces intermittent, hard-to-reproduce corruption under load rather than a clean crash.
Cricket analogy: Single instance with Single concurrency is like a stadium with exactly one turnstile that admits fans one at a time — perfectly orderly but painfully slow for a full house, whereas Single instance with Multiple concurrency is like opening ten turnstiles into the same single stand, which is faster but now needs stewards actively managing the crowd flow to avoid a crush.
Throughput Implications
The practical throughput ceiling of a WCF service is a direct function of these two settings combined with the service throttling behavior covered elsewhere: PerCall with Multiple concurrency (its effective default combination) scales best for stateless workloads since WCF can freely spin up many concurrent instances handled by the thread pool, while Single with Single concurrency caps the service at strictly one operation executing at any given moment regardless of how many CPU cores or how large the thread pool is. A common production mistake is choosing InstanceContextMode.Single for convenience — because it feels like a familiar singleton pattern from other application designs — without realizing that pairing it with the default ConcurrencyMode.Single silently turns a supposedly scalable web service into a de facto single-threaded bottleneck under real concurrent load.
Cricket analogy: It's like scheduling a franchise's entire net-bowling program through one single bowling machine when there are actually six practice lanes available — a well-intentioned simplicity choice that quietly caps every batter's practice throughput to whatever one machine can deliver.
InstanceContextMode.Single combined with the default ConcurrencyMode.Single is a common accidental performance trap: it looks like a normal singleton service but effectively serializes every client call across the entire application, regardless of hardware. If you need a shared singleton for state but also need real throughput, you must explicitly set ConcurrencyMode.Multiple and take full responsibility for thread-safe access to every shared field.
- InstanceContextMode.PerCall creates a fresh, disposable service instance for every call; ideal for stateless, scalable services.
- InstanceContextMode.PerSession keeps one instance alive per client session, enabling conversational state across multiple calls.
- InstanceContextMode.Single shares exactly one instance across the entire application and every client.
- ConcurrencyMode.Single serializes all access to an instance, providing free thread safety at the cost of throughput.
- ConcurrencyMode.Multiple allows concurrent execution inside an instance but requires the developer to manually protect shared state.
- ConcurrencyMode.Reentrant releases the serialization lock during outbound calls to prevent callback deadlocks, at the cost of reintroduced concurrency risk during that window.
- Single instancing with the default Single concurrency silently caps a service to one operation executing at a time — a common accidental production bottleneck.
Practice what you learned
1. What is the key characteristic of InstanceContextMode.PerCall?
2. What does ConcurrencyMode.Single guarantee?
3. What extra responsibility does ConcurrencyMode.Multiple place on the developer?
4. What is a common accidental performance trap involving InstanceContextMode.Single?
5. What does ConcurrencyMode.Reentrant do differently from Single?
Was this page helpful?
You May Also Like
WCF Throttling
Learn how ServiceThrottlingBehavior caps concurrent calls, sessions, and instances to protect a WCF service from being overwhelmed under load.
Faults and Exception Handling
Learn how WCF translates .NET exceptions into SOAP faults, how to design typed FaultContracts, and how to centralize fault handling with IErrorHandler.
WCF Reliable Messaging
Learn how WCF's WS-ReliableMessaging support guarantees ordered, exactly-once (or at-least-once) message delivery over unreliable transports like HTTP.
Related Reading
Related Study Notes in Microsoft Technologies
Browse all study notesWindows 10 / UWP Development Study Notes
.NET · 30 topics
Microsoft TechnologiesWindows Batch Scripting Study Notes
Batch · 30 topics
Microsoft TechnologiesMFC (Microsoft Foundation Classes) Study Notes
C++ · 30 topics
Microsoft TechnologiesSilverlight Study Notes
.NET · 30 topics
Microsoft TechnologiesXAML Study Notes
.NET · 30 topics
Microsoft TechnologiesWPF (Windows Presentation Foundation) Study Notes
.NET · 30 topics