Soenneker.Sets.Concurrent.Bounded 4.0.34

Soenneker.Sets.Concurrent.Bounded

A high-performance bounded concurrent set for .NET

Soenneker.Sets.Concurrent.Bounded provides a thread-safe concurrent set with an approximate maximum size.
It behaves similarly to a ConcurrentDictionary-backed set but automatically evicts older entries when the set grows beyond a configured limit.

Designed for high-throughput workloads where preventing unbounded memory growth is more important than strict eviction guarantees.

Typical uses include:

• deduplication caches
• request ID tracking
• preventing duplicate messages
• bounded concurrent caches
• rate limiting helpers
• abuse protection
• event stream deduplication

Installation

dotnet add package Soenneker.Sets.Concurrent.Bounded

Why this library exists

The .NET ecosystem lacks a simple bounded concurrent set.

Existing options have tradeoffs:

BoundedConcurrentSet provides a lightweight alternative optimized for:

• high concurrency
• predictable memory usage
• low allocation
• minimal locking

Features

✔ High-throughput concurrent operations ✔ Approximate size bounding ✔ Opportunistic eviction of older entries ✔ Lock-minimized design ✔ Low allocation footprint ✔ Safe for heavy multi-threaded workloads

Example

using Soenneker.Sets.Concurrent.Bounded;

var set = new BoundedConcurrentSet<string>(maxSize: 1000);

set.TryAdd("alpha");
set.TryAdd("beta");

bool exists = set.Contains("alpha");

set.TryRemove("beta");

string[] snapshot = set.ToArray();

Configuration

var set = new BoundedConcurrentSet<string>(
    maxSize: 1000,
    capacityHint: 1000,
    trimBatchSize: 64,
    trimStartOveragePercent: 5
);

Performance Characteristics

Eviction work is bounded per call, preventing long latency spikes.

Internally the structure uses:

• concurrent dictionary indexing
• opportunistic FIFO-like eviction
• generation tokens to avoid stale scans
• atomic counters for approximate size tracking

When to use this

Use this collection when you need:

• a bounded concurrent set
• deduplication tracking
• a lightweight concurrent cache
• to prevent unbounded memory growth

Good examples:

• deduplicating phone numbers
• tracking recent request IDs
• preventing duplicate events
• guarding against repeated API calls

Thread Safety

All operations are fully thread-safe.

The implementation is designed for high-concurrency environments and does not require external locking.