👁 Image
Soenneker.Sets.Concurrent.SlidingWindow 4.0.46

👁 alternate text is missing from this package README image
👁 alternate text is missing from this package README image
👁 alternate text is missing from this package README image
👁 alternate text is missing from this package README image

👁 alternate text is missing from this package README image
Soenneker.Sets.Concurrent.SlidingWindow

A high-throughput, thread-safe set whose bucketed entries automatically expire after a fixed time window.

Soenneker.Sets.Concurrent.SlidingWindow provides a concurrent sliding-window set for .NET. Items added to the set automatically expire after a configurable time window without requiring manual cleanup.

The implementation is optimized for high-concurrency workloads and avoids expensive per-item timers by using a bucketed time-slice rotation system.

This makes it ideal for deduplication, rate limiting, and recent activity tracking.

Installation

dotnet add package Soenneker.Sets.Concurrent.SlidingWindow

Why this library exists

Many systems need to track items that should only exist for a limited period of time, such as:

• recently processed messages
• request IDs
• phone numbers
• event IDs
• authentication tokens

Traditional options have downsides:

SlidingWindowConcurrentSet solves this by using bucketed time slices where items automatically expire when their time window passes.

Key Features

✔ High-throughput concurrent operations

✔ Automatic expiration of entries

✔ Sliding window time-based retention

✔ Lock-minimized design

✔ Low allocation footprint

✔ No per-item timers

✔ Safe for heavy multi-threaded workloads

Example

using Soenneker.Sets.Concurrent.SlidingWindow;

var set = new SlidingWindowConcurrentSet<string>(
 window: TimeSpan.FromMinutes(5),
 rotationInterval: TimeSpan.FromSeconds(30)
);

set.TryAdd("alpha");

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

await Task.Delay(TimeSpan.FromMinutes(6));

bool expired = set.Contains("alpha"); // false

Configuration

var set = new SlidingWindowConcurrentSet<string>(
 window: TimeSpan.FromMinutes(10),
 rotationInterval: TimeSpan.FromSeconds(15)
);

The window is internally divided into time buckets:

window / rotationInterval = number of buckets

Example:

window = 5 minutes
rotationInterval = 30 seconds
bucket count = 10

Each rotation advances the window and expires the oldest bucket.

How it works

The set maintains:

• a ConcurrentDictionary for fast lookup
• a ring buffer of queues representing time buckets

When an item is added:

1. It is assigned the current time bucket
2. The value is queued in that bucket
3. The dictionary records the bucket index

A background rotation process periodically:

1. Advances the active bucket
2. Processes the expired bucket
3. Removes items whose last activity falls outside the sliding window

This avoids scanning the entire collection.

Performance Characteristics

The design ensures:

• predictable memory usage
• minimal locking
• bounded cleanup work

Thread Safety

All operations are thread-safe.

The set is designed for high-concurrency environments and does not require external synchronization.

Disposal

The set uses an internal rotation task driven by PeriodicTimer.

When the set is no longer needed, it should be disposed:

set.Dispose();

or

await set.DisposeAsync();

This stops the internal rotation loop.