{"id":49135,"date":"2023-11-28T10:05:00","date_gmt":"2023-11-28T18:05:00","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/dotnet\/?p=49135"},"modified":"2024-12-13T14:03:14","modified_gmt":"2024-12-13T22:03:14","slug":"building-resilient-cloud-services-with-dotnet-8","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/dotnet\/building-resilient-cloud-services-with-dotnet-8\/","title":{"rendered":"Building resilient cloud services with .NET 8"},"content":{"rendered":"

Building resilient apps is a fundamental requirement for cloud development. With .NET 8, we’ve made substantial advancements to simplify the integration of resilience into your applications. We’re excited to introduce the Microsoft.Extensions.Http.Resilience<\/code><\/a> and Microsoft.Extensions.Resilience<\/code><\/a> packages to the .NET ecosystem. These new libraries are based on the Polly<\/a> library, a widely recognized open-source project.<\/p>\n

TL;DR<\/h2>\n

To use the new HTTP resilience APIs, install the package from the command line:<\/p>\n

dotnet add package Microsoft.Extensions.Http.Resilience<\/code><\/pre>\n
Or add it directly in the C# project file:<\/p>\n
<ItemGroup>\r\n  <PackageReference Include=\"Microsoft.Extensions.Http.Resilience\" \/>\r\n<\/ItemGroup><\/code><\/pre>\n
You can now use the recommended AddStandardResilienceHandler<\/code> extension on the IHttpClientBuilder<\/code>:<\/p>\n
var services = new ServiceCollection();\r\n\r\nservices.AddHttpClient(\"my-client\")\r\n        .AddStandardResilienceHandler(options =>\r\n        {\r\n            \/\/ Configure standard resilience options here\r\n        });<\/code><\/pre>\n
Example above uses AddStandardResilienceHandler<\/code> to add a pipeline (rate limiter, total timeout, retry, circuit breaker, attempt timeout) of resilience strategies to the HTTP client. See standard resilience pipeline<\/a> section to learn more.<\/p>\n
A more real-world example would rely on hosting, such as that described in the .NET Generic Host<\/a> article. Using the Microsoft.Extensions.Hosting<\/a> NuGet package, the above example becomes:<\/p>\n
using Http.Resilience.Example;\r\nusing Microsoft.Extensions.DependencyInjection;\r\nusing Microsoft.Extensions.Hosting;\r\n\r\nHostApplicationBuilder builder = Host.CreateApplicationBuilder(args);\r\nIServiceCollection services = builder.Services;\r\n\r\nservices.AddHttpClient(\"my-client\")\r\n        .AddStandardResilienceHandler(options =>\r\n        {\r\n            \/\/ Configure standard resilience options here\r\n        });\r\n\r\n\/\/ Use the client\r\nvar host = builder.Build();\r\nvar httpClient = host.Services\r\n    .GetRequiredService<IHttpClientFactory>()\r\n    .CreateClient(\"my-client\");\r\n\r\n\/\/ Make resilient HTTP request\r\nHttpResponseMessage response = await httpClient.GetAsync(\"https:\/\/jsonplaceholder.typicode.com\/comments\");<\/code><\/pre>\n
To use the registered my-client<\/code>, resolve IHttpClientFactory<\/code> from dependency injection and use the CreateClient<\/code> method to create an HTTP client.<\/p><\/blockquote>\n
For advanced scenarios, the APIs support building your own custom HTTP resilience pipeline:<\/p>\n
services.AddHttpClient(\"my-client\")\r\n        .AddResilienceHandler(\"my-pipeline\", builder =>\r\n        {\r\n            \/\/ Refer to https:\/\/www.pollydocs.org\/strategies\/retry.html#defaults for retry defaults\r\n            builder.AddRetry(new HttpRetryStrategyOptions\r\n            {\r\n                MaxRetryAttempts = 4,\r\n                Delay = TimeSpan.FromSeconds(2),\r\n                BackoffType = DelayBackoffType.Exponential\r\n            });\r\n\r\n            \/\/ Refer to https:\/\/www.pollydocs.org\/strategies\/timeout.html#defaults for timeout defaults\r\n            builder.AddTimeout(TimeSpan.FromSeconds(5));\r\n        });<\/code><\/pre>\n
Why resilience is important<\/h2>\n
Services often rely on the HTTP protocol to make remote requests. These requests can occasionally fail due to network issues or problems on the server side. If not addressed properly, these failures can impact the service’s availability. As your service integrates more remote dependencies, the likelihood of cascading failures, where failures in one system trigger failures in others, increases. Learn more about cascading failures<\/a>.<\/p>\n
To bolster the resilience of your service, especially concerning outgoing HTTP requests, you can employ several strategies:<\/p>\n