\u0130\u015f S\u00fcresi<\/th>\n	Yakla\u015f\u0131m<\/th>\n<\/tr>\n<\/thead>\n
< 2 sn<\/td>\n	Senkron HTTP<\/td>\n<\/tr>\n
2 – 10 sn<\/td>\n	Duruma g\u00f6re senkron veya asenkron s\u00fcre\u00e7<\/td>\n<\/tr>\n
10 sn+<\/td>\n	Background processing<\/td>\n<\/tr>\n
Dakikalar<\/td>\n	Queue + Worker<\/td>\n<\/tr>\n
Saatler<\/td>\n	Workflow \/ Job scheduler<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n Hadi \u015fimdi gelin tek tek desenleri do\u011frusuyla ve yanl\u0131\u015f\u0131yla incelemeye ba\u015flayal\u0131m.<\/p>\n \n Fire-and-Forget<\/em><\/strong> \n\u0130lk olarak bu g\u00fcne kadar g\u00f6rd\u00fc\u011f\u00fcm en tehlikeli yakla\u015f\u0131m\u0131 ele alarak ba\u015flayal\u0131m istiyorum. <\/p>\n \n \r\napp.MapGet("\/fire-and-forget", () =>\r\n{\r\n _ = Task.Run(async () =>\r\n {\r\n while (true)\r\n {\r\n await Task.Delay(1500);\r\n Console.WriteLine("PDF generating...");\r\n }\r\n });\r\n\r\n return TypedResults.Ok();\r\n});\r\n<\/pre>\n<\/div>\n Bu yakla\u015f\u0131mda mant\u0131k; yap\u0131lacak i\u015flem ne kadar uzun olursa olsun istek an\u0131nda sonu\u00e7lansa da i\u015flemin arka planda s\u00fcrd\u00fcr\u00fclmesini sa\u011flamakt\u0131r. \u0130lk bak\u0131\u015fta her ne kadar pratik g\u00f6r\u00fclse de Task.Run<\/code> komutu Asp.NET Core’da g\u00fcvenilir bir arka plan i\u015fleme mekanizmas\u0131 olmad\u0131\u011f\u0131ndan dolay\u0131 esas\u0131nda olduk\u00e7a s\u0131k\u0131nt\u0131l\u0131 bir yakla\u015f\u0131m sergilenmektedir! \u00d6zellikle uygulama yeniden ba\u015flat\u0131l\u0131rsa task yar\u0131da kalabilir yahut hata al\u0131n\u0131rsa kimse fark etmeksizin i\u015flem s\u00fcreci baltalanabilir. Ve t\u00fcm bu s\u00fcre\u00e7te retry gibi bir mekanizma olmayaca\u011f\u0131 i\u00e7in de ciddi kay\u0131plar s\u00f6z konusu olabilir.<\/p>\n `Yan\u0131t s\u00fcresi ile \u00e7al\u0131\u015fma s\u00fcresi ayn\u0131 \u015fey olmak zorunda de\u011fildir! Haliyle request’lerin s\u00fcreleri bu mant\u0131\u011fa g\u00f6re yap\u0131land\u0131r\u0131lmak istenebilir. Ancak yanl\u0131\u015f y\u00f6ntemler ciddi maliyetlere sebebiyet verebilece\u011fi i\u00e7in dikkat edilmelidir…<\/em><\/p><\/blockquote>\n<\/li>\n` Background Queue<\/em><\/strong> \nS\u0131rada Microsoft’un \u00f6nerdi\u011fi yakla\u015f\u0131mdan bahsedelim…<\/p>\nBu yakla\u015f\u0131mda ise yo\u011fun i\u015flem hacmine sahip endpoint’e gelen iste\u011fin direkt i\u015flevsellikten \u00f6nce bir queue’ya aktar\u0131lmas\u0131 esas te\u015fkil etmektedir. \u015e\u00f6yle ki;<\/p>\n \n\n Request\n <\/div>\n\u279c<\/div>\n\nAPI<\/div>\n\n \u0130\u015f kuyru\u011fa eklenir\n <\/div>\n<\/p><\/div>\n\u279c<\/div>\n\n Response\n <\/div>\n\u279c<\/div>\n\nBackground Service<\/div>\n\n Kuyruktan i\u015f al\u0131n\u0131r ve i\u015flenir\n <\/div>\n<\/p><\/div>\n<\/div>\nBurada g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere yine gelen request an\u0131nda response d\u00f6nd\u00fcrecektir. Ancak i\u015f bir \u00f6nceki yakla\u015f\u0131ma nazaran daha g\u00fcvenli bir \u015fekilde queue \u00fczerinden background service ile y\u00fcr\u00fct\u00fcl\u00fcyor olacakt\u0131r. Bunun i\u00e7in uygulaman\u0131n belle\u011finde (in-memory) \u00e7al\u0131\u015fan bir queue’dan yararlan\u0131lmaktad\u0131r. Bak\u0131n dikkat ederseniz hen\u00fcz RabbitMQ<\/a>, NATS<\/a> vs. gibi harici bir message broker altyap\u0131s\u0131na gitmiyoruz! Daha yap\u0131lacak i\u015f hacmi o derece a\u011f\u0131r ve \u00f6l\u00e7eklendirme gerektirmedi\u011fi i\u00e7in i\u015fi in-memory’de \u00e7\u00f6z\u00fcyoruz \ud83d\ude09<\/p>\n Bu yakla\u015f\u0131m sayesinde HTTP request hem uzun s\u00fcre a\u00e7\u0131k kalmayacak hem de kullan\u0131c\u0131 ve client bekletilmeyecektir. Gelen taleplere kar\u015f\u0131n i\u015fler, s\u0131ras\u0131yla veya belirlendi\u011fi say\u0131da paralel bir \u015fekilde i\u015fleniyor olacakt\u0131r. Task.Run<\/code> yakla\u015f\u0131m\u0131na g\u00f6re \u00e7ok daha kontroll\u00fc ve emniyetlidir! Ancak \u00f6nemli bir dezavantaj\u0131 vard\u0131r ki, o da, queue’nun uygulaman\u0131n belle\u011finde olmas\u0131d\u0131r. Uygulama yeniden ba\u015flat\u0131ld\u0131\u011f\u0131 taktirde queue’daki i\u015fler kaybolacakt\u0131r. Bu nedenle kritik sistemlerde kesinlikle harici message broker kullan\u0131lmas\u0131 elzem olacakt\u0131r.<\/p>\n \u00d6rnek \u00e7al\u0131\u015fma;<\/p>\n \n\r\napp.MapGet("\/background-queue", async () =>\r\n{\r\n await BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n {\r\n for (int i = 0; i < 10; i++)\r\n {\r\n await Task.Delay(1500, cancellationToken);\r\n Console.WriteLine($"PDF generating... {i}");\r\n }\r\n });\r\n\r\n return TypedResults.Ok();\r\n});\r\n<\/pre>\n<\/div>\nBurada gelen istek neticesinde yap\u0131lacak i\u015f (her ne kadar for d\u00f6ng\u00fcs\u00fc de olsa) in-memory’de arka planda i\u015flenir halde queue’ya al\u0131nmaktad\u0131r.<\/p>\n\n \u26a0\ufe0f<\/strong> Bu yakla\u015f\u0131m\u0131n kod b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc g\u00f6rebilmek i\u00e7in \u015furadaki<\/a> \u00e7al\u0131\u015fmaya g\u00f6z atabilirsiniz.\n<\/div>\n<\/li>\nMessage Broker<\/em><\/strong> \n\u0130\u015f b\u00fcy\u00fcd\u00fc\u011f\u00fcnde, queue art\u0131k process i\u00e7erisinden \u00e7\u0131kar\u0131lmal\u0131d\u0131r. Bu art\u0131k ger\u00e7ek anlamda \u00f6l\u00e7eklenebilir bir yap\u0131 sunacakt\u0131r.<\/p>\nBu yakla\u015f\u0131ma dair \u00f6nceden klavyeye ald\u0131\u011f\u0131m a\u015fa\u011f\u0131daki i\u00e7eriklere g\u00f6z atman\u0131z\u0131 tavsiye ederim:<\/p>\n\n.NET\u2019te NATS JetStream \u0130le \u00c7al\u0131\u015fmak<\/a><\/li>\n Asp.NET Core \u2013 MassTransit & RabbitMQ \u0130le Birlikte Messaging Uygulamas\u0131<\/a><\/li>\n<\/ul>\n<\/li>\n Polling Pattern<\/em><\/strong> \nBu y\u00f6ntem de y\u0131llard\u0131r kullan\u0131lan en g\u00fcvenilir yakla\u015f\u0131mlardan biridir. \u0130\u015fin ba\u015flat\u0131lmas\u0131, ama sonucunun beklenmemesi \u00fczerine kurulmu\u015f bir davran\u0131\u015fa sahiptir. Sonucu daha sonra gelip sorulmal\u0131d\u0131r.<\/p>\nTabi gelip neticenin sorulabilmesi i\u00e7in bir jobId<\/em> de\u011ferine ihtiya\u00e7 olacakt\u0131r. Bunun i\u00e7in API, gelen request neticesinde an\u0131nda <\/p>\n{ \n “jobId”: “a8f3d1” \n}\n<\/div>\ntarz\u0131 bir jobId<\/em> de\u011feri d\u00f6nd\u00fcrecektir.<\/p>\n Neden buna polling denmektedir?<\/em><\/strong> \n\u00c7\u00fcnk\u00fc client belirli aral\u0131klarla sunucuya ‘Bitti mi?’ diye soru sormaktad\u0131r. Yani ileti\u015fim client taraf\u0131ndad\u0131r. Sunucu, i\u015f bitince kullan\u0131c\u0131ya kendili\u011finden haber vermemektedir!<\/p>\n\nClient \n \u2502 \nPOST \/reports \n \u2502 \n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 JobId \n \u2502 \nGET \/reports\/{id} \n \u2502 \n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Running \n \u2502 \nGET \/reports\/{id} \n \u2502 \n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Running \n \u2502 \nGET \/reports\/{id} \n \u2502 \n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Completed\n<\/div>\nBu yakla\u015f\u0131m\u0131 \u015f\u00f6yle bir \u00e7al\u0131\u015fma \u00fczerinden \u00f6rneklendirebiliriz;<\/p>\n\n\r\napp.MapPost("\/polling-pattern", async (JobStore jobStore) =>\r\n{\r\n var jobInfo = jobStore.Create();\r\n await AspNetCore.ApiScalingPatterns.PollingPattern.BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n {\r\n jobInfo.Status = JobStatus.Running;\r\n jobStore.Update(jobInfo);\r\n\r\n \/\/\u0130\u015eLEM YAPILIYOR\r\n try\r\n {\r\n for (int i = 0; i < 10; i++)\r\n {\r\n await Task.Delay(1500, cancellationToken);\r\n Console.WriteLine($"PDF generating... {i}");\r\n }\r\n jobInfo.Status = JobStatus.Completed;\r\n jobInfo.ResultUrl = "https:\/\/example.com\/result.pdf";\r\n }\r\n catch (Exception ex)\r\n {\r\n jobInfo.Status = JobStatus.Failed;\r\n jobInfo.Error = ex.Message;\r\n }\r\n \/\/\u0130\u015eLEM YAPILIYOR\r\n\r\n jobStore.Update(jobInfo);\r\n });\r\n\r\n return TypedResults.Ok(new\r\n {\r\n jobId = jobInfo.Id\r\n });\r\n});\r\n<\/pre>\n<\/div>\n\n\r\napp.MapGet("\/polling-pattern\/{jobId}", async (JobStore jobStore, Guid jobId) =>\r\n{\r\n var jobInfo = jobStore.Get(jobId);\r\n\r\n return TypedResults.Ok(jobInfo);\r\n});\r\n<\/pre>\n<\/div>\nEvet, burada g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere ikinci olu\u015fturdu\u011fumuz GET t\u00fcr\u00fcnden endpoint sayesinde job’a dair durum sorgulamas\u0131n\u0131 yapabilir ve yap\u0131lacak i\u015fe dair kontrolleri sa\u011flayabiliriz. <\/p>\n\n \u26a0\ufe0f<\/strong> Bu yakla\u015f\u0131m\u0131n kod b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc g\u00f6rebilmek i\u00e7in \u015furadaki<\/a> \u00e7al\u0131\u015fmaya g\u00f6z atabilirsiniz.\n<\/div>\n<\/li>\nCallback \/ Webhook<\/em><\/strong> \nSen beni bekleme, i\u015fim bitince ben seri arar\u0131m…<\/em><\/p><\/blockquote>\n Polling’de client s\u00fcrekli sunucuya sorarken, Webhook’ta sunucu i\u015flem bittikten sonra client’a haber vermektedir. Yap\u0131land\u0131rmas\u0131 olduk\u00e7a basit mant\u0131\u011fa dayanmaktad\u0131r. Hemen \u00f6rneklendirirsek e\u011fer;<\/p>\n\n\r\napp.MapPost("\/callback-webhook\/{*callbackUrl}", async (string callbackUrl, JobStore jobStore, IHttpClientFactory httpClientFactory) =>\r\n{\r\n var jobId = Guid.NewGuid();\r\n\r\n await AspNetCore.ApiScalingPatterns.Callback_Webhook.BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n {\r\n for (int i = 0; i < 10; i++)\r\n {\r\n await Task.Delay(1500, cancellationToken);\r\n Console.WriteLine($"PDF generating... {i}");\r\n }\r\n\r\n var httpClient = httpClientFactory.CreateClient();\r\n await httpClient.PostAsJsonAsync(\r\n requestUri: callbackUrl,\r\n value: new\r\n {\r\n JobId = jobId,\r\n Status = "Completed",\r\n DownloadUrl = "https:\/\/example.com\/result.pdf"\r\n }\r\n );\r\n });\r\n\r\n return TypedResults.Ok(new\r\n {\r\n jobId = jobId\r\n });\r\n});\r\n<\/pre>\n<\/div>\nYukar\u0131daki \u00e7al\u0131\u015fmay\u0131 incelerseniz e\u011fer uzun s\u00fcre vakit alacak olan request’in i\u015flemini JobStore<\/em>‘a yazmak yerine bir callback i\u015flemi e\u015fli\u011finde yine bir queue’ya almaktay\u0131z. \u0130\u015flem bitti\u011fi taktirde g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere callback tetiklenecektir. Bu callback endpoint’ini de a\u015fa\u011f\u0131daki gibi tasarlayabiliriz:<\/p>\n\n\r\napp.MapPost("\/api\/report-completed", (ReportCompletedRequest request, ILogger<Program> logger) =>\r\n{\r\n Console.WriteLine($"{request.JobId} \| {request.Status} \| {request.DownloadUrl}");\r\n return Results.Ok();\r\n});\r\n<\/pre>\n<\/div>\n<\/li>\nWorkflow Engine<\/em><\/strong> \nBaz\u0131 i\u015fler ise tek bir metottan ibaret olmamaktad\u0131r. Haliyle gelen request neticesinde birden fazla metot, hatta d\u0131\u015f servis ve hatta distributed bir sistem tetiklenmeye ba\u015flayabilir. \u0130\u015fte b\u00f6yle durumlar i\u00e7in queue tek ba\u015f\u0131na yeterli olmayacakt\u0131r. Burada devreye workflow yakla\u015f\u0131m\u0131 girebilir. Bunun i\u00e7in sizlere \u00f6nerebilece\u011fim en \u00e7a\u011fda\u015f ve etkin workflow teknolojisi Dapr Workflow<\/a>‘dur. Bu yakla\u015f\u0131m sayesinde sistemin state’ini rahatl\u0131kla tutabilir, retry’\u0131 kolayca yap\u0131land\u0131rabilirsiniz. Ayr\u0131ca ihtiya\u00e7 do\u011frultusunda compensation uygulayabilir, timeout y\u00f6netimi sa\u011flayabilir ve t\u00fcm bu s\u00fcreci durable execution ile sa\u011fl\u0131kl\u0131 bir \u015fekilde y\u00f6netebilirsiniz.\n<\/li>\n Streaming<\/em><\/strong> \nYo\u011fun ve uzun da olsa baz\u0131 request’lerin neticelerini beklemek i\u015f mant\u0131\u011f\u0131 gere\u011fi zaruri olabilmektedir. B\u00f6yle durumlarda gelen request’in neticesini stream bir \u015fekilde response edebiliriz. Bunun i\u00e7in Server-Sent Events (SSE)<\/a> veya gRPC Streaming<\/a> gibi \u00e7\u00f6z\u00fcmlere gidilebilir. Bu \u00e7\u00f6z\u00fcmlerde connection s\u00fcrekli a\u00e7\u0131k kalacak ancak client veya kullan\u0131c\u0131 verisel i\u015fleyi\u015f bak\u0131m\u0131ndan ilerlemeyi hissediyor olacakt\u0131r.\n<\/li>\n<\/ul>\nBurada inceledi\u011fimiz y\u00f6ntemlerde ele ald\u0131\u011f\u0131m\u0131z bellek i\u00e7i (in-memory) kuyruk (queue) yakla\u015f\u0131m\u0131na ek olarak, bir veritaban\u0131 tabanl\u0131 \u00e7\u00f6z\u00fcmden de yararlan\u0131labilir. Esas olan, belirlenen amaca ula\u015fmakt\u0131r. Bu yakla\u015f\u0131mlar\u0131n temel hedefi, \u00f6zellikle endpoint’lerde meydana gelebilecek ani y\u00fck art\u0131\u015flar\u0131 kar\u015f\u0131s\u0131nda sistemi daha dayan\u0131kl\u0131 ve g\u00fcvenilir h\u00e2le getirmektir.<\/p>\n K\u0131saca \u00f6zetlersek;<\/p>\n\nGelen request’ler i\u00e7erisinde yava\u015f ya da uzun vadeli i\u015flerin optimize edilmesi gerekmektedir. Bir \u015fekilde bunlar\u0131n arka planda ya da farkl\u0131 bir ba\u011flamda \u00e7\u00f6z\u00fclmesi gerekmektedir.<\/li>\n Endpoint i\u00e7erisinde Task.Run<\/code> gibi ka\u00e7\u0131\u015f y\u00f6ntemlerinden uzak durulmal\u0131d\u0131r. Gerekirse i\u015flem bir queue’ya al\u0131nmal\u0131d\u0131r ama ekstradan bir thread olu\u015fturmaktan ka\u00e7\u0131n\u0131lmal\u0131d\u0131r!<\/li>\n\u0130\u015flem tamamland\u0131\u011f\u0131nda ise kullan\u0131c\u0131 yahut client bilgilendirilmelidir! Evet, belirli periyotta i\u015flemin durumunu yoklama y\u00f6ntemiyle kontrol etmek iyi olabilir, ancak anl\u0131k bildirim daha iyidir!<\/li>\n<\/ul>\n\u0130lgilenenlerin faydalanmas\u0131 dile\u011fiyle… \nSonraki yaz\u0131lar\u0131mda g\u00f6r\u00fc\u015fmek \u00fczere… \n\u0130yi \u00e7al\u0131\u015fmalar…<\/p>\n\n\n \n\u00d6rnek \u00e7al\u0131\u015fmaya a\u015fa\u011f\u0131daki GitHub reposundan eri\u015febilirsiniz.<\/div>\nhttps:\/\/github.com\/gncyyldz\/AspNetCore.ApiScalingPatterns<\/div>\n<\/p><\/div>\n<\/p><\/div>\n\n Bu repository, ilgili konuya dair \u00f6rnek \u00e7al\u0131\u015fman\u0131n kaynak kodlar\u0131n\u0131 ve mimari yap\u0131s\u0131n\u0131 i\u00e7ermektedir. Detaylar i\u00e7in GitHub \u00fczerinden incelemede bulunabilirsiniz.\n <\/div>\n \n GitHub\u2019da G\u00f6r\u00fcnt\u00fcle \u2192 \n <\/a><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":" Merhaba, Geli\u015ftirdi\u011fimiz uygulamalar\u0131m\u0131zda illa ki tamamlanmas\u0131 dakikalarca s\u00fcren bir endpoint’in ortaya \u00e7\u0131kmas\u0131 neredeyse ka\u00e7\u0131n\u0131lmazd\u0131r. Misal olarak; y\u0131llarca toplanan verilerin istatistiki olarak raporlanmas\u0131 gerekti\u011fi durumlarda, toplu veri aktar\u0131mlar\u0131nda veya response d\u00f6nd\u00fcr\u00fclmeden \u00f6nce birka\u00e7 servise ve...<\/p>\n","protected":false},"author":1,"featured_media":28859,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5206,10,3102],"tags":[2679],"class_list":["post-28850","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-net","category-asp-net","category-asp-net-core","tag-asp-net-core"],"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/comments?post=28850"}],"version-history":[{"count":9,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850\/revisions"}],"predecessor-version":[{"id":28861,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850\/revisions\/28861"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/media\/28859"}],"wp:attachment":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/media?parent=28850"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/categories?post=28850"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/tags?post=28850"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

\r\napp.MapGet("\/fire-and-forget", () =>\r\n{\r\n    _ = Task.Run(async () =>\r\n    {\r\n        while (true)\r\n        {\r\n            await Task.Delay(1500);\r\n            Console.WriteLine("PDF generating...");\r\n        }\r\n    });\r\n\r\n    return TypedResults.Ok();\r\n});\r\n<\/pre>\n<\/div>\nBu yakla\u015f\u0131mda mant\u0131k; yap\u0131lacak i\u015flem ne kadar uzun olursa olsun istek an\u0131nda sonu\u00e7lansa da i\u015flemin arka planda s\u00fcrd\u00fcr\u00fclmesini sa\u011flamakt\u0131r. \u0130lk bak\u0131\u015fta her ne kadar pratik g\u00f6r\u00fclse de Task.Run<\/code> komutu Asp.NET Core’da g\u00fcvenilir bir arka plan i\u015fleme mekanizmas\u0131 olmad\u0131\u011f\u0131ndan dolay\u0131 esas\u0131nda olduk\u00e7a s\u0131k\u0131nt\u0131l\u0131 bir yakla\u015f\u0131m sergilenmektedir! \u00d6zellikle uygulama yeniden ba\u015flat\u0131l\u0131rsa task yar\u0131da kalabilir yahut hata al\u0131n\u0131rsa kimse fark etmeksizin i\u015flem s\u00fcreci baltalanabilir. Ve t\u00fcm bu s\u00fcre\u00e7te retry gibi bir mekanizma olmayaca\u011f\u0131 i\u00e7in de ciddi kay\u0131plar s\u00f6z konusu olabilir.<\/p>\n
Yan\u0131t s\u00fcresi ile \u00e7al\u0131\u015fma s\u00fcresi ayn\u0131 \u015fey olmak zorunda de\u011fildir! Haliyle request’lerin s\u00fcreleri bu mant\u0131\u011fa g\u00f6re yap\u0131land\u0131r\u0131lmak istenebilir. Ancak yanl\u0131\u015f y\u00f6ntemler ciddi maliyetlere sebebiyet verebilece\u011fi i\u00e7in dikkat edilmelidir…<\/em><\/p><\/blockquote>\n<\/li>\n
Background Queue<\/em><\/strong>
\nS\u0131rada Microsoft’un \u00f6nerdi\u011fi yakla\u015f\u0131mdan bahsedelim…<\/p>\nBu yakla\u015f\u0131mda ise yo\u011fun i\u015flem hacmine sahip endpoint’e gelen iste\u011fin direkt i\u015flevsellikten \u00f6nce bir queue’ya aktar\u0131lmas\u0131 esas te\u015fkil etmektedir. \u015e\u00f6yle ki;<\/p>\n
\n\n        Request\n    <\/div>\n\u279c<\/div>\n\nAPI<\/div>\n\n            \u0130\u015f kuyru\u011fa eklenir\n        <\/div>\n<\/p><\/div>\n\u279c<\/div>\n\n        Response\n    <\/div>\n\u279c<\/div>\n\nBackground Service<\/div>\n\n            Kuyruktan i\u015f al\u0131n\u0131r ve i\u015flenir\n        <\/div>\n<\/p><\/div>\n<\/div>\nBurada g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere yine gelen request an\u0131nda response d\u00f6nd\u00fcrecektir. Ancak i\u015f bir \u00f6nceki yakla\u015f\u0131ma nazaran daha g\u00fcvenli bir \u015fekilde queue \u00fczerinden background service ile y\u00fcr\u00fct\u00fcl\u00fcyor olacakt\u0131r. Bunun i\u00e7in uygulaman\u0131n belle\u011finde (in-memory) \u00e7al\u0131\u015fan bir queue’dan yararlan\u0131lmaktad\u0131r. Bak\u0131n dikkat ederseniz hen\u00fcz RabbitMQ<\/a>, NATS<\/a> vs. gibi harici bir message broker altyap\u0131s\u0131na gitmiyoruz! Daha yap\u0131lacak i\u015f hacmi o derece a\u011f\u0131r ve \u00f6l\u00e7eklendirme gerektirmedi\u011fi i\u00e7in i\u015fi in-memory’de \u00e7\u00f6z\u00fcyoruz \ud83d\ude09<\/p>\n
Bu yakla\u015f\u0131m sayesinde HTTP request hem uzun s\u00fcre a\u00e7\u0131k kalmayacak hem de kullan\u0131c\u0131 ve client bekletilmeyecektir. Gelen taleplere kar\u015f\u0131n i\u015fler, s\u0131ras\u0131yla veya belirlendi\u011fi say\u0131da paralel bir \u015fekilde i\u015fleniyor olacakt\u0131r. Task.Run<\/code> yakla\u015f\u0131m\u0131na g\u00f6re \u00e7ok daha kontroll\u00fc ve emniyetlidir! Ancak \u00f6nemli bir dezavantaj\u0131 vard\u0131r ki, o da, queue’nun uygulaman\u0131n belle\u011finde olmas\u0131d\u0131r. Uygulama yeniden ba\u015flat\u0131ld\u0131\u011f\u0131 taktirde queue’daki i\u015fler kaybolacakt\u0131r. Bu nedenle kritik sistemlerde kesinlikle harici message broker kullan\u0131lmas\u0131 elzem olacakt\u0131r.<\/p>\n
\u00d6rnek \u00e7al\u0131\u015fma;<\/p>\n
\n\r\napp.MapGet("\/background-queue", async () =>\r\n{\r\n    await BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n      {\r\n          for (int i = 0; i < 10; i++)\r\n          {\r\n              await Task.Delay(1500, cancellationToken);\r\n              Console.WriteLine($"PDF generating... {i}");\r\n          }\r\n      });\r\n\r\n    return TypedResults.Ok();\r\n});\r\n<\/pre>\n<\/div>\nBurada gelen istek neticesinde yap\u0131lacak i\u015f (her ne kadar for d\u00f6ng\u00fcs\u00fc de olsa) in-memory’de arka planda i\u015flenir halde queue’ya al\u0131nmaktad\u0131r.<\/p>\n\n    \u26a0\ufe0f<\/strong> Bu yakla\u015f\u0131m\u0131n kod b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc g\u00f6rebilmek i\u00e7in \u015furadaki<\/a> \u00e7al\u0131\u015fmaya g\u00f6z atabilirsiniz.\n<\/div>\n<\/li>\nMessage Broker<\/em><\/strong>
\n\u0130\u015f b\u00fcy\u00fcd\u00fc\u011f\u00fcnde, queue art\u0131k process i\u00e7erisinden \u00e7\u0131kar\u0131lmal\u0131d\u0131r. Bu art\u0131k ger\u00e7ek anlamda \u00f6l\u00e7eklenebilir bir yap\u0131 sunacakt\u0131r.<\/p>\nBu yakla\u015f\u0131ma dair \u00f6nceden klavyeye ald\u0131\u011f\u0131m a\u015fa\u011f\u0131daki i\u00e7eriklere g\u00f6z atman\u0131z\u0131 tavsiye ederim:<\/p>\n\n.NET\u2019te NATS JetStream \u0130le \u00c7al\u0131\u015fmak<\/a><\/li>\n
Asp.NET Core \u2013 MassTransit & RabbitMQ \u0130le Birlikte Messaging Uygulamas\u0131<\/a><\/li>\n<\/ul>\n<\/li>\n
Polling Pattern<\/em><\/strong>
\nBu y\u00f6ntem de y\u0131llard\u0131r kullan\u0131lan en g\u00fcvenilir yakla\u015f\u0131mlardan biridir. \u0130\u015fin ba\u015flat\u0131lmas\u0131, ama sonucunun beklenmemesi \u00fczerine kurulmu\u015f bir davran\u0131\u015fa sahiptir. Sonucu daha sonra gelip sorulmal\u0131d\u0131r.<\/p>\nTabi gelip neticenin sorulabilmesi i\u00e7in bir jobId<\/em> de\u011ferine ihtiya\u00e7 olacakt\u0131r. Bunun i\u00e7in API, gelen request neticesinde an\u0131nda <\/p>\n{
\n    “jobId”: “a8f3d1”
\n}\n<\/div>\ntarz\u0131 bir jobId<\/em> de\u011feri d\u00f6nd\u00fcrecektir.<\/p>\n
Neden buna polling denmektedir?<\/em><\/strong>
\n\u00c7\u00fcnk\u00fc client belirli aral\u0131klarla sunucuya ‘Bitti mi?’ diye soru sormaktad\u0131r. Yani ileti\u015fim client taraf\u0131ndad\u0131r. Sunucu, i\u015f bitince kullan\u0131c\u0131ya kendili\u011finden haber vermemektedir!<\/p>\n\nClient
\n   \u2502
\nPOST \/reports
\n   \u2502
\n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 JobId
\n   \u2502
\nGET \/reports\/{id}
\n   \u2502
\n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Running
\n   \u2502
\nGET \/reports\/{id}
\n   \u2502
\n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Running
\n   \u2502
\nGET \/reports\/{id}
\n   \u2502
\n\u25c4\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500 Completed\n<\/div>\nBu yakla\u015f\u0131m\u0131 \u015f\u00f6yle bir \u00e7al\u0131\u015fma \u00fczerinden \u00f6rneklendirebiliriz;<\/p>\n\n\r\napp.MapPost("\/polling-pattern", async (JobStore jobStore) =>\r\n{\r\n    var jobInfo = jobStore.Create();\r\n    await AspNetCore.ApiScalingPatterns.PollingPattern.BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n    {\r\n        jobInfo.Status = JobStatus.Running;\r\n        jobStore.Update(jobInfo);\r\n\r\n        \/\/\u0130\u015eLEM YAPILIYOR\r\n        try\r\n        {\r\n            for (int i = 0; i < 10; i++)\r\n            {\r\n                await Task.Delay(1500, cancellationToken);\r\n                Console.WriteLine($"PDF generating... {i}");\r\n            }\r\n            jobInfo.Status = JobStatus.Completed;\r\n            jobInfo.ResultUrl = "https:\/\/example.com\/result.pdf";\r\n        }\r\n        catch (Exception ex)\r\n        {\r\n            jobInfo.Status = JobStatus.Failed;\r\n            jobInfo.Error = ex.Message;\r\n        }\r\n        \/\/\u0130\u015eLEM YAPILIYOR\r\n\r\n        jobStore.Update(jobInfo);\r\n    });\r\n\r\n    return TypedResults.Ok(new\r\n    {\r\n        jobId = jobInfo.Id\r\n    });\r\n});\r\n<\/pre>\n<\/div>\n\n\r\napp.MapGet("\/polling-pattern\/{jobId}", async (JobStore jobStore, Guid jobId) =>\r\n{\r\n    var jobInfo = jobStore.Get(jobId);\r\n\r\n    return TypedResults.Ok(jobInfo);\r\n});\r\n<\/pre>\n<\/div>\nEvet, burada g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere ikinci olu\u015fturdu\u011fumuz GET t\u00fcr\u00fcnden endpoint sayesinde job’a dair durum sorgulamas\u0131n\u0131 yapabilir ve yap\u0131lacak i\u015fe dair kontrolleri sa\u011flayabiliriz. <\/p>\n\n    \u26a0\ufe0f<\/strong> Bu yakla\u015f\u0131m\u0131n kod b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc g\u00f6rebilmek i\u00e7in \u015furadaki<\/a> \u00e7al\u0131\u015fmaya g\u00f6z atabilirsiniz.\n<\/div>\n<\/li>\nCallback \/ Webhook<\/em><\/strong>
\nSen beni bekleme, i\u015fim bitince ben seri arar\u0131m…<\/em><\/p><\/blockquote>\n
Polling’de client s\u00fcrekli sunucuya sorarken, Webhook’ta sunucu i\u015flem bittikten sonra client’a haber vermektedir. Yap\u0131land\u0131rmas\u0131 olduk\u00e7a basit mant\u0131\u011fa dayanmaktad\u0131r. Hemen \u00f6rneklendirirsek e\u011fer;<\/p>\n\n\r\napp.MapPost("\/callback-webhook\/{*callbackUrl}", async (string callbackUrl, JobStore jobStore, IHttpClientFactory httpClientFactory) =>\r\n{\r\n    var jobId = Guid.NewGuid();\r\n\r\n    await AspNetCore.ApiScalingPatterns.Callback_Webhook.BackgroundTaskQueue.EnqueueAsync(async cancellationToken =>\r\n    {\r\n        for (int i = 0; i < 10; i++)\r\n        {\r\n            await Task.Delay(1500, cancellationToken);\r\n            Console.WriteLine($"PDF generating... {i}");\r\n        }\r\n\r\n        var httpClient = httpClientFactory.CreateClient();\r\n        await httpClient.PostAsJsonAsync(\r\n            requestUri: callbackUrl,\r\n            value: new\r\n            {\r\n                JobId = jobId,\r\n                Status = "Completed",\r\n                DownloadUrl = "https:\/\/example.com\/result.pdf"\r\n            }\r\n        );\r\n    });\r\n\r\n    return TypedResults.Ok(new\r\n    {\r\n        jobId = jobId\r\n    });\r\n});\r\n<\/pre>\n<\/div>\nYukar\u0131daki \u00e7al\u0131\u015fmay\u0131 incelerseniz e\u011fer uzun s\u00fcre vakit alacak olan request’in i\u015flemini JobStore<\/em>‘a yazmak yerine bir callback i\u015flemi e\u015fli\u011finde yine bir queue’ya almaktay\u0131z. \u0130\u015flem bitti\u011fi taktirde g\u00f6r\u00fcld\u00fc\u011f\u00fc \u00fczere callback tetiklenecektir. Bu callback endpoint’ini de a\u015fa\u011f\u0131daki gibi tasarlayabiliriz:<\/p>\n\n\r\napp.MapPost("\/api\/report-completed", (ReportCompletedRequest request, ILogger<Program> logger) =>\r\n{\r\n    Console.WriteLine($"{request.JobId} | {request.Status} | {request.DownloadUrl}");\r\n    return Results.Ok();\r\n});\r\n<\/pre>\n<\/div>\n<\/li>\nWorkflow Engine<\/em><\/strong>
\nBaz\u0131 i\u015fler ise tek bir metottan ibaret olmamaktad\u0131r. Haliyle gelen request neticesinde birden fazla metot, hatta d\u0131\u015f servis ve hatta distributed bir sistem tetiklenmeye ba\u015flayabilir. \u0130\u015fte b\u00f6yle durumlar i\u00e7in queue tek ba\u015f\u0131na yeterli olmayacakt\u0131r. Burada devreye workflow yakla\u015f\u0131m\u0131 girebilir. Bunun i\u00e7in sizlere \u00f6nerebilece\u011fim en \u00e7a\u011fda\u015f ve etkin workflow teknolojisi Dapr Workflow<\/a>‘dur. Bu yakla\u015f\u0131m sayesinde sistemin state’ini rahatl\u0131kla tutabilir, retry’\u0131 kolayca yap\u0131land\u0131rabilirsiniz. Ayr\u0131ca ihtiya\u00e7 do\u011frultusunda compensation uygulayabilir, timeout y\u00f6netimi sa\u011flayabilir ve t\u00fcm bu s\u00fcreci durable execution ile sa\u011fl\u0131kl\u0131 bir \u015fekilde y\u00f6netebilirsiniz.\n<\/li>\n
Streaming<\/em><\/strong>
\nYo\u011fun ve uzun da olsa baz\u0131 request’lerin neticelerini beklemek i\u015f mant\u0131\u011f\u0131 gere\u011fi zaruri olabilmektedir. B\u00f6yle durumlarda gelen request’in neticesini stream bir \u015fekilde response edebiliriz. Bunun i\u00e7in Server-Sent Events (SSE)<\/a> veya gRPC Streaming<\/a> gibi \u00e7\u00f6z\u00fcmlere gidilebilir. Bu \u00e7\u00f6z\u00fcmlerde connection s\u00fcrekli a\u00e7\u0131k kalacak ancak client veya kullan\u0131c\u0131 verisel i\u015fleyi\u015f bak\u0131m\u0131ndan ilerlemeyi hissediyor olacakt\u0131r.\n<\/li>\n<\/ul>\nBurada inceledi\u011fimiz y\u00f6ntemlerde ele ald\u0131\u011f\u0131m\u0131z bellek i\u00e7i (in-memory) kuyruk (queue) yakla\u015f\u0131m\u0131na ek olarak, bir veritaban\u0131 tabanl\u0131 \u00e7\u00f6z\u00fcmden de yararlan\u0131labilir. Esas olan, belirlenen amaca ula\u015fmakt\u0131r. Bu yakla\u015f\u0131mlar\u0131n temel hedefi, \u00f6zellikle endpoint’lerde meydana gelebilecek ani y\u00fck art\u0131\u015flar\u0131 kar\u015f\u0131s\u0131nda sistemi daha dayan\u0131kl\u0131 ve g\u00fcvenilir h\u00e2le getirmektir.<\/p>\n
K\u0131saca \u00f6zetlersek;<\/p>\n\nGelen request’ler i\u00e7erisinde yava\u015f ya da uzun vadeli i\u015flerin optimize edilmesi gerekmektedir. Bir \u015fekilde bunlar\u0131n arka planda ya da farkl\u0131 bir ba\u011flamda \u00e7\u00f6z\u00fclmesi gerekmektedir.<\/li>\n
Endpoint i\u00e7erisinde Task.Run<\/code> gibi ka\u00e7\u0131\u015f y\u00f6ntemlerinden uzak durulmal\u0131d\u0131r. Gerekirse i\u015flem bir queue’ya al\u0131nmal\u0131d\u0131r ama ekstradan bir thread olu\u015fturmaktan ka\u00e7\u0131n\u0131lmal\u0131d\u0131r!<\/li>\n\u0130\u015flem tamamland\u0131\u011f\u0131nda ise kullan\u0131c\u0131 yahut client bilgilendirilmelidir! Evet, belirli periyotta i\u015flemin durumunu yoklama y\u00f6ntemiyle kontrol etmek iyi olabilir, ancak anl\u0131k bildirim daha iyidir!<\/li>\n<\/ul>\n\u0130lgilenenlerin faydalanmas\u0131 dile\u011fiyle…
\nSonraki yaz\u0131lar\u0131mda g\u00f6r\u00fc\u015fmek \u00fczere…
\n\u0130yi \u00e7al\u0131\u015fmalar…<\/p>\n\n\n    \n\u00d6rnek \u00e7al\u0131\u015fmaya a\u015fa\u011f\u0131daki GitHub reposundan eri\u015febilirsiniz.<\/div>\nhttps:\/\/github.com\/gncyyldz\/AspNetCore.ApiScalingPatterns<\/div>\n<\/p><\/div>\n<\/p><\/div>\n\n    Bu repository, ilgili konuya dair \u00f6rnek \u00e7al\u0131\u015fman\u0131n kaynak kodlar\u0131n\u0131 ve mimari yap\u0131s\u0131n\u0131 i\u00e7ermektedir. Detaylar i\u00e7in GitHub \u00fczerinden incelemede bulunabilirsiniz.\n  <\/div>\n  
\n    GitHub\u2019da G\u00f6r\u00fcnt\u00fcle \u2192
\n  <\/a><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"
Merhaba, Geli\u015ftirdi\u011fimiz uygulamalar\u0131m\u0131zda illa ki tamamlanmas\u0131 dakikalarca s\u00fcren bir endpoint’in ortaya \u00e7\u0131kmas\u0131 neredeyse ka\u00e7\u0131n\u0131lmazd\u0131r. Misal olarak; y\u0131llarca toplanan verilerin istatistiki olarak raporlanmas\u0131 gerekti\u011fi durumlarda, toplu veri aktar\u0131mlar\u0131nda veya response d\u00f6nd\u00fcr\u00fclmeden \u00f6nce birka\u00e7 servise ve...<\/p>\n","protected":false},"author":1,"featured_media":28859,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5206,10,3102],"tags":[2679],"class_list":["post-28850","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-net","category-asp-net","category-asp-net-core","tag-asp-net-core"],"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/comments?post=28850"}],"version-history":[{"count":9,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850\/revisions"}],"predecessor-version":[{"id":28861,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/posts\/28850\/revisions\/28861"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/media\/28859"}],"wp:attachment":[{"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/media?parent=28850"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/categories?post=28850"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.gencayyildiz.com\/blog\/wp-json\/wp\/v2\/tags?post=28850"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}