IAsyncEnumerable<T>

Overview

IAsyncEnumerable<T> is the async counterpart of IEnumerable<T>, enabling pull-based streaming of data where each element is produced asynchronously. It uses yield return inside async iterator methods and is consumed with await foreach. Unlike Task<List<T>> which waits for all items before returning, IAsyncEnumerable<T> yields items one at a time as they become available, reducing memory pressure and time-to-first-result for large datasets.

IAsyncEnumerable<T> is supported natively in ASP.NET Core (controller actions, minimal APIs), Entity Framework Core (streaming query results), gRPC (server streaming), and System.Text.Json (streaming serialization). It was introduced in C# 8 / .NET Core 3.0 and requires no additional packages.

Basic Producer and Consumer

csharp

using System.Collections.Generic;
using System.Runtime.CompilerServices;
using System.Threading;
using System.Threading.Tasks;

namespace MyApp.Services;

public class SensorDataService
{
    // Producer: yields sensor readings one at a time
    public async IAsyncEnumerable<SensorReading> StreamReadingsAsync(
        string sensorId,
        [EnumeratorCancellation] CancellationToken cancellationToken = default)
    {
        while (!cancellationToken.IsCancellationRequested)
        {
            var reading = await ReadSensorAsync(sensorId, cancellationToken);
            yield return reading;
            await Task.Delay(TimeSpan.FromSeconds(1), cancellationToken);
        }
    }

    private async Task<SensorReading> ReadSensorAsync(
        string sensorId, CancellationToken ct)
    {
        // Simulate async I/O
        await Task.Delay(10, ct);
        return new SensorReading(
            SensorId: sensorId,
            Value: Random.Shared.NextDouble() * 100,
            Timestamp: DateTime.UtcNow);
    }
}

public record SensorReading(string SensorId, double Value, DateTime Timestamp);

csharp

using System;
using System.Threading;
using System.Threading.Tasks;
using MyApp.Services;

// Consumer: processes items as they arrive
public class SensorMonitor
{
    private readonly SensorDataService _service;

    public SensorMonitor(SensorDataService service)
    {
        _service = service;
    }

    public async Task MonitorAsync(string sensorId, CancellationToken ct)
    {
        await foreach (var reading in _service.StreamReadingsAsync(sensorId, ct))
        {
            if (reading.Value > 90.0)
            {
                Console.WriteLine(
                    $"ALERT: Sensor {reading.SensorId} reading {reading.Value:F1} " +
                    $"at {reading.Timestamp:HH:mm:ss}");
            }
        }
    }
}

Entity Framework Core Streaming

EF Core's AsAsyncEnumerable() streams query results without loading all rows into memory.

csharp

using System.Collections.Generic;
using System.Linq;
using System.Runtime.CompilerServices;
using System.Threading;
using Microsoft.EntityFrameworkCore;

namespace MyApp.Data;

public class OrderRepository
{
    private readonly AppDbContext _context;

    public OrderRepository(AppDbContext context)
    {
        _context = context;
    }

    // Streams orders from the database without loading all into memory
    public async IAsyncEnumerable<Order> GetLargeOrdersAsync(
        decimal minTotal,
        [EnumeratorCancellation] CancellationToken cancellationToken = default)
    {
        await foreach (var order in _context.Orders
            .Where(o => o.Total >= minTotal)
            .OrderByDescending(o => o.CreatedAt)
            .AsAsyncEnumerable()
            .WithCancellation(cancellationToken))
        {
            yield return order;
        }
    }

    // Batch processing with streaming
    public async IAsyncEnumerable<Order[]> GetOrderBatchesAsync(
        int batchSize,
        [EnumeratorCancellation] CancellationToken cancellationToken = default)
    {
        var batch = new List<Order>(batchSize);

        await foreach (var order in _context.Orders
            .AsAsyncEnumerable()
            .WithCancellation(cancellationToken))
        {
            batch.Add(order);
            if (batch.Count >= batchSize)
            {
                yield return batch.ToArray();
                batch.Clear();
            }
        }

        if (batch.Count > 0)
        {
            yield return batch.ToArray();
        }
    }
}

public class Order
{
    public int Id { get; set; }
    public decimal Total { get; set; }
    public DateTime CreatedAt { get; set; }
}

public class AppDbContext : DbContext
{
    public DbSet<Order> Orders => Set<Order>();

    public AppDbContext(DbContextOptions<AppDbContext> options) : base(options) { }
}

ASP.NET Core Integration

ASP.NET Core natively supports IAsyncEnumerable<T> in controller actions and minimal APIs, streaming JSON array elements as they are produced.

csharp

using System.Collections.Generic;
using System.Runtime.CompilerServices;
using System.Threading;
using Microsoft.AspNetCore.Mvc;
using MyApp.Data;

namespace MyApp.Controllers;

[ApiController]
[Route("api/[controller]")]
public class OrdersController : ControllerBase
{
    private readonly OrderRepository _repository;

    public OrdersController(OrderRepository repository)
    {
        _repository = repository;
    }

    // ASP.NET Core streams the JSON array elements one at a time
    [HttpGet("large")]
    public IAsyncEnumerable<Order> GetLargeOrders(
        [FromQuery] decimal minTotal = 1000,
        CancellationToken cancellationToken = default)
    {
        return _repository.GetLargeOrdersAsync(minTotal, cancellationToken);
    }
}

csharp

// Minimal API equivalent
app.MapGet("/api/orders/stream", (
    OrderRepository repo,
    CancellationToken ct) =>
{
    return repo.GetLargeOrdersAsync(1000, ct);
});

LINQ Operators for IAsyncEnumerable

System.Linq.Async (NuGet package) provides LINQ operators for IAsyncEnumerable<T>.

csharp

using System.Collections.Generic;
using System.Linq;
using System.Runtime.CompilerServices;
using System.Threading;
using System.Threading.Tasks;

namespace MyApp.Services;

public class AnalyticsService
{
    private readonly SensorDataService _sensors;

    public AnalyticsService(SensorDataService sensors)
    {
        _sensors = sensors;
    }

    // Filter, transform, and take from an async stream
    public async IAsyncEnumerable<Alert> GetAlertsAsync(
        string sensorId,
        [EnumeratorCancellation] CancellationToken ct = default)
    {
        await foreach (var alert in _sensors.StreamReadingsAsync(sensorId, ct)
            .Where(r => r.Value > 85.0)
            .Select(r => new Alert(
                r.SensorId,
                $"High reading: {r.Value:F1}",
                r.Timestamp))
            .Take(50)
            .WithCancellation(ct))
        {
            yield return alert;
        }
    }

    // Aggregate over a stream
    public async Task<double> GetAverageAsync(
        string sensorId, int sampleCount, CancellationToken ct)
    {
        return await _sensors.StreamReadingsAsync(sensorId, ct)
            .Take(sampleCount)
            .AverageAsync(r => r.Value, ct);
    }

    // Buffer into chunks
    public async IAsyncEnumerable<SensorReading[]> GetBufferedAsync(
        string sensorId,
        int bufferSize,
        [EnumeratorCancellation] CancellationToken ct = default)
    {
        await foreach (var chunk in _sensors.StreamReadingsAsync(sensorId, ct)
            .Buffer(bufferSize)
            .WithCancellation(ct))
        {
            yield return chunk.ToArray();
        }
    }
}

public record Alert(string SensorId, string Message, DateTime Timestamp);

IAsyncEnumerable vs. Alternatives

Feature	IAsyncEnumerable<T>	Task<List<T>>	IObservable<T>	Channel<T>
Pull/Push model	Pull (consumer controls)	Pull (all at once)	Push (producer controls)	Push with backpressure
Memory usage	O(1) per item	O(n) all items	O(1) per item	O(capacity)
Cancellation	CancellationToken	CancellationToken	Dispose subscription	CancellationToken
Backpressure	Natural (consumer speed)	N/A	Must implement manually	Bounded channel
Multiple consumers	Not supported natively	Share the list	Multiple subscriptions	Multiple readers
LINQ support	System.Linq.Async	System.Linq	System.Reactive.Linq	None built-in
Best for	Streaming query results	Small result sets	Event streams, UI events	Producer/consumer queues

Best Practices

•
Always accept [EnumeratorCancellation] CancellationToken as the last parameter of async iterator methods so that consumers can cancel enumeration; the attribute wires the token from WithCancellation() to the parameter automatically.
•
Use await foreach with ConfigureAwait(false) in library code by writing await foreach (var item in stream.ConfigureAwait(false)) to avoid capturing the synchronization context on each iteration.
•
Return IAsyncEnumerable<T> from ASP.NET Core endpoints for large result sets instead of Task<List<T>> so that ASP.NET streams JSON array elements incrementally, reducing memory allocation and time-to-first-byte.
•
Use AsAsyncEnumerable() in EF Core queries instead of ToListAsync() for large datasets to stream rows from the database one at a time; note that the DbContext must remain alive for the duration of the enumeration.
•
Install the System.Linq.Async NuGet package for LINQ operators (Where, Select, Take, Skip, AverageAsync, Buffer) because the BCL does not include LINQ extension methods for IAsyncEnumerable<T>.
•
Do not enumerate the same IAsyncEnumerable<T> instance multiple times because each enumeration restarts the producer; if multiple consumers need the data, materialize it into a list first or use a different abstraction like channels.
•
Use yield return with try/finally for resource cleanup instead of manual iterator state machines, because the compiler generates the correct disposal logic when await using or await foreach exits (including on cancellation or exception).
•
Prefer IAsyncEnumerable<T> over IObservable<T> when the consumer controls the pace (pull-based) -- database queries, file reading, paginated APIs -- and prefer IObservable<T> when the producer controls the pace (push-based) -- UI events, message bus subscriptions.
•
Batch items using Buffer(count) from System.Linq.Async when processing items individually is too slow -- for example, inserting 100 rows at a time into a database instead of one at a time, reducing round-trip overhead.
•
Test async iterators by collecting results into a list with await stream.ToListAsync() or by asserting on individual items with await foreach and a counter, ensuring the stream terminates correctly and cancellation is honored.