MongoDB, ilk piyasaya s\u00fcr\u00fcld\u00fc\u011f\u00fcnde esnek \u015fema yap\u0131s\u0131, y\u00fcksek performansl\u0131 okuma\/yazma yetenekleri ve yatay \u00f6l\u00e7eklenebilirlik vaatleriyle dikkat \u00e7ekmi\u015f, “ACID” (Atomicity, Consistency, Isolation, Durability) \u00f6zelliklerini tam olarak desteklemeyen bir NoSQL veritaban\u0131 olarak konumlanm\u0131\u015ft\u0131r. Geleneksel ili\u015fkisel veritabanlar\u0131n\u0131n aksine, MongoDB’nin temel felsefesi, verilerin tek bir belge i\u00e7inde atomik olarak i\u015flenmesi ve da\u011f\u0131t\u0131k sistemlerde “nihai tutarl\u0131l\u0131k” (eventual consistency) ilkesi \u00fczerine kuruluydu. Ancak modern uygulama gereksinimleri, \u00f6zellikle finans, e-ticaret gibi alanlarda, birden fazla belge ve\/veya koleksiyon \u00fczerinde tam atomiklik ve g\u00fc\u00e7l\u00fc tutarl\u0131l\u0131k talep etmeye ba\u015flad\u0131. Bu ihtiyaca yan\u0131t olarak, MongoDB 4.0 s\u00fcr\u00fcm\u00fcyle birlikte da\u011f\u0131t\u0131k i\u015flemler (distributed transactions) deste\u011fini tan\u0131tt\u0131 ve 4.2 s\u00fcr\u00fcm\u00fcyle sharded cluster’larda da tam destek sa\u011flayarak bu yetene\u011fi daha da geli\u015ftirdi.<\/p>\n
Veritaban\u0131 sistemlerinde i\u015flemler, bir dizi operasyonun tek bir atomik birim olarak y\u00fcr\u00fct\u00fclmesini sa\u011flayan kritik bir mekanizmad\u0131r. Bu, ya t\u00fcm operasyonlar\u0131n ba\u015far\u0131yla tamamlanmas\u0131 ya da hi\u00e7birinin tamamlanmamas\u0131 gerekti\u011fi anlam\u0131na gelir. E\u011fer bu operasyonlardan herhangi biri ba\u015far\u0131s\u0131z olursa, sistem \u00f6nceki kararl\u0131 durumuna geri d\u00f6ner. Bu “hepsi ya da hi\u00e7biri” ilkesi, veri b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc ve tutarl\u0131l\u0131\u011f\u0131n\u0131 sa\u011flamak i\u00e7in vazge\u00e7ilmezdir.<\/p>\n
Geleneksel ili\u015fkisel veritabanlar\u0131, ACID \u00f6zelliklerini temelden destekleyerek y\u0131llard\u0131r bu ihtiyac\u0131 kar\u015f\u0131lam\u0131\u015ft\u0131r. MongoDB gibi NoSQL veritabanlar\u0131, genellikle \u00f6l\u00e7eklenebilirlik ve performans u\u011fruna bu kat\u0131 ACID garantilerinden baz\u0131lar\u0131n\u0131 feda etmi\u015ftir. MongoDB’nin belge odakl\u0131 yap\u0131s\u0131, tek bir belge i\u00e7indeki operasyonlar\u0131n atomik olmas\u0131n\u0131 sa\u011flarken, birden fazla belge veya koleksiyonu kapsayan operasyonlar i\u00e7in atomiklik garantisi sunmuyordu. \u00d6rne\u011fin, bir banka uygulamas\u0131nda bir hesaptan di\u011ferine para transferi yaparken, hem g\u00f6nderen hesab\u0131n bakiyesinin azalt\u0131lmas\u0131 hem de al\u0131c\u0131 hesab\u0131n bakiyesinin art\u0131r\u0131lmas\u0131 gerekir. Bu iki operasyonun atomik olarak ger\u00e7ekle\u015fmesi zorunludur. E\u011fer g\u00f6nderen hesab\u0131n bakiyesi azalt\u0131l\u0131r, ancak al\u0131c\u0131n\u0131n bakiyesi art\u0131r\u0131lamazsa (\u00f6rne\u011fin bir sistem hatas\u0131 nedeniyle), veritaban\u0131 tutars\u0131z bir duruma d\u00fc\u015fer ve para kaybolmu\u015f gibi g\u00f6r\u00fcn\u00fcr.<\/p>\n Da\u011f\u0131t\u0131k sistemlerde bu t\u00fcr senaryolar\u0131 y\u00f6netmek, \u00f6zellikle a\u011f gecikmeleri, d\u00fc\u011f\u00fcm ar\u0131zalar\u0131 ve e\u015fzamanl\u0131 eri\u015fim gibi fakt\u00f6rler nedeniyle daha da karma\u015f\u0131k hale gelir. MongoDB’nin i\u015flemleri desteklemeye ba\u015flamas\u0131, geli\u015ftiricilere da\u011f\u0131t\u0131k ortamlarda bile g\u00fc\u00e7l\u00fc veri tutarl\u0131l\u0131\u011f\u0131 garantileriyle uygulamalar olu\u015fturma esnekli\u011fi sa\u011flam\u0131\u015ft\u0131r. Bu, MongoDB’nin kullan\u0131m alanlar\u0131n\u0131 geni\u015fletmi\u015f ve onu hem esneklik hem de tutarl\u0131l\u0131k gerektiren kritik i\u015f y\u00fckleri i\u00e7in daha uygun bir se\u00e7enek haline getirmi\u015ftir.<\/p>\n MongoDB’nin 4.0 s\u00fcr\u00fcm\u00fcyle tan\u0131t\u0131lan i\u015flemler, \u00e7oklu belge, \u00e7oklu koleksiyon ve hatta \u00e7oklu veritaban\u0131 operasyonlar\u0131n\u0131 tek bir atomik birim olarak y\u00fcr\u00fctme yetene\u011fi sunar. Bu, daha \u00f6nce karma\u015f\u0131k uygulama seviyesi mant\u0131k gerektiren veya m\u00fcmk\u00fcn olmayan senaryolar\u0131 basitle\u015ftirir ve g\u00fcvenilir hale getirir.<\/p>\n MongoDB i\u015flemleri, da\u011f\u0131t\u0131k ortamlarda tam ACID garantileri sunar: Ba\u015flang\u0131\u00e7ta MongoDB 4.0 i\u015flemleri yaln\u0131zca replika setleri \u00fczerinde destekleniyordu. MongoDB 4.2 ile birlikte, i\u015flemler sharded cluster’lar \u00fczerinde de tam olarak desteklenmeye ba\u015fland\u0131. Bu, MongoDB’nin b\u00fcy\u00fck \u00f6l\u00e7ekli, da\u011f\u0131t\u0131k mimarilerde bile veri tutarl\u0131l\u0131\u011f\u0131 gerektiren uygulamalar i\u00e7in eksiksiz bir \u00e7\u00f6z\u00fcm sundu\u011fu anlam\u0131na gelir.<\/p>\n \u0130\u015flemler i\u00e7inde yap\u0131lan okuma ve yazma operasyonlar\u0131, i\u015flemin kendisi i\u00e7in tan\u0131mlanm\u0131\u015f MongoDB i\u015flemleri, tek bir i\u015flem i\u00e7inde 16 MB’a kadar veri boyutunu ve maksimum 1000 operasyonu destekler. Bu s\u0131n\u0131rlar, \u00e7o\u011fu uygulama senaryosu i\u00e7in yeterli olsa da, \u00e7ok b\u00fcy\u00fck veya uzun s\u00fcreli i\u015flemlerden ka\u00e7\u0131nmak en iyi uygulamad\u0131r.<\/p>\n MongoDB i\u015flemleri, veri tutarl\u0131l\u0131\u011f\u0131n\u0131n kritik oldu\u011fu ve birden fazla belge veya koleksiyonun ayn\u0131 anda g\u00fcncellenmesi gereken bir\u00e7ok senaryoda kullan\u0131l\u0131r:<\/p>\n * Banka ve Finans Uygulamalar\u0131:<\/strong> * E-ticaret ve Stok Y\u00f6netimi:<\/strong> * Envanter ve Lojistik Sistemleri:<\/strong> * \u00c7ok Ad\u0131ml\u0131 Veri G\u00fcncellemeleri:<\/strong> * Muhasebe Sistemleri:<\/strong> Bu senaryolar\u0131n her birinde, i\u015flemlerin kullan\u0131lmas\u0131, uygulaman\u0131n veri b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc ve g\u00fcvenilirli\u011fini \u00f6nemli \u00f6l\u00e7\u00fcde art\u0131r\u0131r. Geli\u015ftiriciler, daha \u00f6nce uygulama katman\u0131nda manuel olarak y\u00f6netmek zorunda kald\u0131klar\u0131 karma\u015f\u0131k tutarl\u0131l\u0131k mant\u0131\u011f\u0131n\u0131 art\u0131k veritaban\u0131 seviyesinde, daha g\u00fcvenli ve standart bir \u015fekilde ele alabilirler.<\/p>\n MongoDB i\u015flemleri, s\u00fcr\u00fcc\u00fcler (drivers) arac\u0131l\u0131\u011f\u0131yla kullan\u0131l\u0131r. \u00c7o\u011fu MongoDB s\u00fcr\u00fcc\u00fcs\u00fc, i\u015flemleri y\u00f6netmek i\u00e7in benzer bir API sa\u011flar. A\u015fa\u011f\u0131daki \u00f6rneklerde Node.js s\u00fcr\u00fcc\u00fcs\u00fc kullan\u0131larak temel bir i\u015flem ak\u0131\u015f\u0131 ve hata y\u00f6netimi g\u00f6sterilecektir.<\/p>\n Bir i\u015flem ba\u015flatmak ve y\u00f6netmek i\u00e7in genellikle \u015fu ad\u0131mlar izlenir: Da\u011f\u0131t\u0131k sistemlerde, a\u011f hatalar\u0131, sunucu yeniden ba\u015flatmalar\u0131 veya ge\u00e7ici kilitlenmeler nedeniyle i\u015flemler ba\u015far\u0131s\u0131z olabilir. MongoDB, bu t\u00fcr ge\u00e7ici hatalar i\u00e7in \u00f6zel hata kodlar\u0131 (\u00f6rne\u011fin MongoDB s\u00fcr\u00fcc\u00fcleri genellikle i\u015flemleri yeniden deneme mant\u0131\u011f\u0131n\u0131 basitle\u015ftirmek i\u00e7in E\u011fer Bu \u00f6rnek, \u0130\u015flem ba\u015flat\u0131l\u0131rken MongoDB i\u015flemleri g\u00fc\u00e7l\u00fc bir ara\u00e7 olsa da, kullan\u0131rken baz\u0131 hususlara ve s\u0131n\u0131rlamalara dikkat etmek \u00f6nemlidir.<\/p>\n \u0130\u015flemler, ek protokol y\u00fck\u00fc, kilitlenme mekanizmalar\u0131 ve replikasyon senkronizasyonu nedeniyle tekil operasyonlara g\u00f6re daha y\u00fcksek gecikme s\u00fcresine sahip olabilir. Bu nedenlerle, i\u015flemleri m\u00fcmk\u00fcn oldu\u011funca k\u0131sa tutmak ve yaln\u0131zca ger\u00e7ekten atomiklik gerektiren durumlar i\u00e7in kullanmak en iyi uygulamad\u0131r.<\/p>\n MongoDB, uzun s\u00fcreli i\u015flemlerin kaynaklar\u0131 t\u00fcketmesini \u00f6nlemek i\u00e7in varsay\u0131lan bir ya\u015fam s\u00fcresi s\u0131n\u0131r\u0131na sahiptir ( MongoDB 4.0 ile i\u015flemler sadece replika setleri i\u00e7inde destekleniyordu. MongoDB 4.2 ile birlikte, sharded cluster’lar \u00fczerinde de \u00e7oklu belge, \u00e7oklu koleksiyon ve \u00e7oklu veritaban\u0131 i\u015flemlerini desteklemeye ba\u015flad\u0131. Bu, sharding kullanan b\u00fcy\u00fck \u00f6l\u00e7ekli uygulamalar i\u00e7in kritik bir geli\u015fmedir. Ancak, sharded cluster’larda i\u015flem performans\u0131 daha karma\u015f\u0131k olabilir ve dikkatli izleme gerektirebilir.<\/p>\n Baz\u0131 veritaban\u0131 operasyonlar\u0131 i\u015flemler i\u00e7inde desteklenmez: Bir belgenin shard anahtar\u0131n\u0131 de\u011fi\u015ftiren g\u00fcncellemeler, bir i\u015flem i\u00e7inde desteklenmez. Bu t\u00fcr operasyonlar, sharding mimarisinin temelini etkiledi\u011fi i\u00e7in i\u015flem d\u0131\u015f\u0131nda yap\u0131lmal\u0131d\u0131r.<\/p>\n MongoDB, tek bir belge \u00fczerindeki t\u00fcm yazma operasyonlar\u0131n\u0131n zaten atomik oldu\u011funu garanti eder. Bu, bir belgenin g\u00fcncellenmesi veya eklenmesi gibi tek belge operasyonlar\u0131 i\u00e7in ayr\u0131ca bir i\u015flem ba\u015flatmaya gerek olmad\u0131\u011f\u0131 anlam\u0131na gelir. \u0130\u015flemler, birden fazla belgeyi veya koleksiyonu etkileyen atomik operasyonlar i\u00e7in tasarlanm\u0131\u015ft\u0131r.<\/p>\n Kulland\u0131\u011f\u0131n\u0131z MongoDB s\u00fcr\u00fcc\u00fcs\u00fcn\u00fcn (\u00f6rne\u011fin Node.js, Python, Java vb.) i\u015flemler \u00f6zelli\u011fini destekleyen g\u00fcncel bir s\u00fcr\u00fcm oldu\u011fundan emin olun. Eski s\u00fcr\u00fcc\u00fcler bu i\u015flevselli\u011fi sa\u011flamayabilir.<\/p>\n MongoDB i\u015flemlerini etkin ve verimli bir \u015fekilde kullanmak i\u00e7in a\u015fa\u011f\u0131daki en iyi uygulamalar\u0131 g\u00f6z \u00f6n\u00fcnde bulundurun:<\/p>\n * \u0130\u015flemleri K\u0131sa Tutun:<\/strong> \u0130\u015flemler, kilitlenmeleri ve performans d\u00fc\u015f\u00fc\u015flerini en aza indirmek i\u00e7in m\u00fcmk\u00fcn oldu\u011funca az operasyon i\u00e7ermeli ve h\u0131zl\u0131 bir \u015fekilde tamamlanmal\u0131d\u0131r. Uzun s\u00fcreli i\u015flemler, di\u011fer operasyonlar\u0131n beklemesine neden olabilir. MongoDB’nin i\u015flem deste\u011fi, veritaban\u0131n\u0131n evriminde \u00f6nemli bir kilometre ta\u015f\u0131d\u0131r. Bu \u00f6zellik, geli\u015ftiricilere da\u011f\u0131t\u0131k sistemlerde dahi g\u00fc\u00e7l\u00fc veri tutarl\u0131l\u0131\u011f\u0131 garantileriyle karma\u015f\u0131k uygulamalar olu\u015fturma imkan\u0131 sunar. \u00d6zellikle finans, e-ticaret ve envanter y\u00f6netimi gibi kritik i\u015f y\u00fcklerinde, birden fazla belge veya koleksiyonu atomik olarak g\u00fcncelleme ihtiyac\u0131 vazge\u00e7ilmezdir.<\/p>\n \u0130\u015flemlerin do\u011fru bir \u015fekilde kullan\u0131lmas\u0131, veri b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc sa\u011flarken, uygulama kodundaki karma\u015f\u0131k tutarl\u0131l\u0131k mant\u0131\u011f\u0131n\u0131 ortadan kald\u0131r\u0131r. Ancak, performans etkileri, desteklenmeyen operasyonlar ve yeniden deneme mekanizmalar\u0131 gibi hususlara dikkat etmek, bu g\u00fc\u00e7l\u00fc \u00f6zelli\u011fi verimli bir \u015fekilde kullanman\u0131n anahtar\u0131d\u0131r. En iyi uygulamalar\u0131 takip ederek ve sisteminizi dikkatle izleyerek, MongoDB i\u015flemlerinin sundu\u011fu avantajlardan tam olarak faydalanabilir ve uygulamalar\u0131n\u0131z\u0131n g\u00fcvenilirli\u011fini \u00f6nemli \u00f6l\u00e7\u00fcde art\u0131rabilirsiniz. MongoDB, esnekli\u011fini ve \u00f6l\u00e7eklenebilirli\u011fini korurken, art\u0131k geleneksel veritabanlar\u0131n\u0131n g\u00fc\u00e7l\u00fc tutarl\u0131l\u0131k garantilerini de sunarak, daha geni\u015f bir kullan\u0131m alan\u0131 yelpazesi i\u00e7in cazip bir se\u00e7enek haline gelmi\u015ftir.<\/body><\/p>\n","protected":false},"excerpt":{"rendered":"MongoDB’de \u0130\u015flemler (Transactions) Nas\u0131l Kullan\u0131l\u0131r?\nMongoDB, ilk piyasaya s\u00fcr\u00fcld\u00fc\u011f\u00fcnde esnek \u015fema yap\u0131s\u0131, y\u00fcksek performansl\u0131 okuma\/y","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"csco_page_header_type":"","csco_page_load_nextpost":"","csco_page_subscribe_form":"","csco_page_contact_form":"","footnotes":""},"categories":[676],"tags":[],"class_list":{"0":"post-31977","1":"post","2":"type-post","3":"status-publish","4":"format-standard","6":"category-mongodb","7":"cs-entry","8":"cs-video-wrap"},"yoast_head":"\n
\n* Atomiklik (Atomicity):<\/strong> Bir i\u015flemdeki t\u00fcm operasyonlar tek bir mant\u0131ksal birim olarak kabul edilir. Ya hepsi ba\u015far\u0131l\u0131 olur ya da hi\u00e7biri.
\n* Tutarl\u0131l\u0131k (Consistency):<\/strong> Bir i\u015flem, veritaban\u0131n\u0131 bir ge\u00e7erli durumdan ba\u015fka bir ge\u00e7erli duruma ta\u015f\u0131r. Veritaban\u0131 k\u0131s\u0131tlamalar\u0131 (constraint) i\u015flem boyunca korunur.
\n* \u0130zolasyon (Isolation):<\/strong> E\u015fzamanl\u0131 y\u00fcr\u00fct\u00fclen i\u015flemler birbirini etkilemez. Her i\u015flem, sanki veritaban\u0131nda tek ba\u015f\u0131na \u00e7al\u0131\u015f\u0131yormu\u015f gibi g\u00f6r\u00fcn\u00fcr.
\n* Dayan\u0131kl\u0131l\u0131k (Durability):<\/strong> Ba\u015far\u0131yla tamamlanan (commit edilen) bir i\u015flemin sonu\u00e7lar\u0131 kal\u0131c\u0131d\u0131r ve sistem \u00e7\u00f6kse bile kaybolmaz.<\/p>\nMongoDB \u0130\u015flemlerine Genel Bak\u0131\u015f<\/h3>\n
ACID \u00d6zellikleri ve MongoDB \u0130\u015flemleri<\/h4>\n
\n* Atomiklik:<\/strong> Bir i\u015flemdeki t\u00fcm operasyonlar ya ba\u015far\u0131l\u0131 olur ya da t\u00fcm\u00fc geri al\u0131n\u0131r.
\n* Tutarl\u0131l\u0131k:<\/strong> \u0130\u015flem, veritaban\u0131n\u0131 her zaman ge\u00e7erli bir duruma ta\u015f\u0131r.
\n* \u0130zolasyon:<\/strong> \u0130\u015flemler, snapshot<\/code> izolasyon seviyesiyle birbirlerinden izole edilir. Bir i\u015flem devam ederken, di\u011fer i\u015flemler taraf\u0131ndan yap\u0131lan de\u011fi\u015fiklikler bu i\u015flem taraf\u0131ndan g\u00f6r\u00fclmez. Yaln\u0131zca i\u015flemin ba\u015flang\u0131c\u0131ndaki veri snapshot’\u0131 \u00fczerinden \u00e7al\u0131\u015f\u0131r. linearizable<\/code> read concern ile daha g\u00fc\u00e7l\u00fc izolasyon da sa\u011flanabilir.
\n* Dayan\u0131kl\u0131l\u0131k:<\/strong> Commit edilen bir i\u015flemin t\u00fcm de\u011fi\u015fiklikleri, replika setindeki \u00e7o\u011fu d\u00fc\u011f\u00fcme (majority) yaz\u0131ld\u0131\u011f\u0131nda kal\u0131c\u0131 hale gelir ve sistem hatalar\u0131ndan etkilenmez.<\/p>\nReplica Set ve Sharded Cluster Deste\u011fi<\/h4>\n
Transaction Write Concern ve Read Concern<\/h4>\n
readConcern<\/code> ve writeConcern<\/code> ayarlar\u0131na tabidir.
\n* writeConcern<\/code>:<\/strong> \u0130\u015flemin commit edilmesi i\u00e7in ka\u00e7 replika set \u00fcyesinin yazma i\u015flemini onaylamas\u0131 gerekti\u011fini belirler. Genellikle majority<\/code> (\u00e7o\u011funluk) olarak ayarlanmas\u0131 \u00f6nerilir, bu da verinin kal\u0131c\u0131l\u0131\u011f\u0131n\u0131 garanti eder.
\n* readConcern<\/code>:<\/strong> \u0130\u015flem i\u00e7indeki okuma operasyonlar\u0131n\u0131n ne kadar g\u00fcncel veya tutarl\u0131 veri g\u00f6rmesi gerekti\u011fini belirler. snapshot<\/code> read concern, i\u015flemin ba\u015flang\u0131c\u0131ndaki verinin anl\u0131k g\u00f6r\u00fcnt\u00fcs\u00fcn\u00fc g\u00f6rmeyi sa\u011flar ve i\u015flem boyunca okuma tutarl\u0131l\u0131\u011f\u0131n\u0131 garanti eder. linearizable<\/code> read concern ise en g\u00fc\u00e7l\u00fc tutarl\u0131l\u0131\u011f\u0131 sunar ancak performans maliyeti daha y\u00fcksektir.<\/p>\n\u0130\u015flem (Transaction) Kullan\u0131m Senaryolar\u0131<\/h3>\n
\n * Para Transferi:<\/strong> Bir hesaptan para \u00e7ekilirken, ayn\u0131 anda ba\u015fka bir hesaba para yat\u0131r\u0131lmas\u0131 gerekir. Bu iki operasyonun atomik olmas\u0131, paran\u0131n kaybolmamas\u0131n\u0131 veya m\u00fckerrer olu\u015fmamas\u0131n\u0131 sa\u011flar.
\n * Kredi Kart\u0131 \u0130\u015flemleri:<\/strong> Bir \u00f6deme al\u0131nd\u0131\u011f\u0131nda, hem m\u00fc\u015fteri bakiyesinin g\u00fcncellenmesi hem de i\u015flem kayd\u0131n\u0131n olu\u015fturulmas\u0131 gibi birden fazla ad\u0131m\u0131n tutarl\u0131 bir \u015fekilde y\u00fcr\u00fct\u00fclmesi.<\/p>\n
\n * Sipari\u015f \u0130\u015fleme:<\/strong> Bir m\u00fc\u015fteri sipari\u015f verdi\u011finde, \u00fcr\u00fcn stoklar\u0131n\u0131n azalt\u0131lmas\u0131, sipari\u015f kayd\u0131n\u0131n olu\u015fturulmas\u0131 ve \u00f6deme durumunun g\u00fcncellenmesi gibi i\u015flemlerin tamam\u0131n\u0131n ba\u015far\u0131l\u0131 olmas\u0131 gerekir. E\u011fer stok azalt\u0131l\u0131r ancak sipari\u015f kaydedilemezse, stok bilgisi yanl\u0131\u015f olur.
\n * Sepet Y\u00f6netimi:<\/strong> Bir kullan\u0131c\u0131n\u0131n sepetindeki \u00fcr\u00fcnlerin g\u00fcncellenmesi veya sat\u0131n alma i\u015flemi s\u0131ras\u0131nda sepetin bo\u015falt\u0131lmas\u0131 ve stoklar\u0131n ayarlanmas\u0131.<\/p>\n
\n * \u00dcr\u00fcn Hareketi:<\/strong> Bir \u00fcr\u00fcn\u00fcn bir depodan di\u011ferine ta\u015f\u0131nmas\u0131 durumunda, her iki deponun envanter kay\u0131tlar\u0131n\u0131n ayn\u0131 anda ve atomik olarak g\u00fcncellenmesi.
\n * \u00dcretim Takibi:<\/strong> Bir \u00fcr\u00fcn\u00fcn \u00fcretim hatt\u0131ndaki farkl\u0131 a\u015famalar\u0131n\u0131 takip ederken, her a\u015fama de\u011fi\u015fikli\u011finin ilgili t\u00fcm kay\u0131tlar\u0131 tutarl\u0131 bir \u015fekilde g\u00fcncellemesi.<\/p>\n
\n * Kullan\u0131c\u0131 Kay\u0131t ve Profil Y\u00f6netimi:<\/strong> Bir kullan\u0131c\u0131n\u0131n kayd\u0131 s\u0131ras\u0131nda hem kullan\u0131c\u0131 belgesinin olu\u015fturulmas\u0131 hem de ilgili rol veya izin belgelerinin g\u00fcncellenmesi.
\n * \u0130\u00e7erik Y\u00f6netim Sistemleri (CMS):<\/strong> Bir blog yaz\u0131s\u0131n\u0131n yay\u0131nlanmas\u0131 s\u0131ras\u0131nda hem yaz\u0131n\u0131n durumunun g\u00fcncellenmesi hem de ilgili etiketlerin veya kategorilerin sayac\u0131n\u0131n art\u0131r\u0131lmas\u0131.<\/p>\n
\n * Defter Kay\u0131tlar\u0131:<\/strong> Bir muhasebe i\u015fleminde, birden fazla defter hesab\u0131n\u0131n ayn\u0131 anda bor\u00e7land\u0131r\u0131lmas\u0131 ve alacakland\u0131r\u0131lmas\u0131.<\/p>\nMongoDB \u0130\u015flemleri Nas\u0131l Kullan\u0131l\u0131r? (Kod \u00d6rnekleri ile)<\/h3>\n
Temel \u0130\u015flem Ak\u0131\u015f\u0131<\/h4>\n
\n1. Bir Client Session ba\u015flat\u0131n:<\/strong> \u0130\u015flemler bir session (oturum) ba\u011flam\u0131nda y\u00fcr\u00fct\u00fcl\u00fcr.
\n2. \u0130\u015flemi ba\u015flat\u0131n:<\/strong> startTransaction()<\/code> metodunu \u00e7a\u011f\u0131rarak i\u015flemi ba\u015flat\u0131n.
\n3. Veritaban\u0131 operasyonlar\u0131n\u0131 ger\u00e7ekle\u015ftirin:<\/strong> Bu operasyonlar session<\/code> nesnesine ba\u011fl\u0131 olmal\u0131d\u0131r.
\n4. \u0130\u015flemi commit edin:<\/strong> T\u00fcm operasyonlar ba\u015far\u0131l\u0131 olursa commitTransaction()<\/code> \u00e7a\u011f\u0131r\u0131n.
\n5. \u0130\u015flemi iptal edin (abort):<\/strong> Herhangi bir hata olu\u015fursa abortTransaction()<\/code> \u00e7a\u011f\u0131r\u0131n.
\n6. Session’\u0131 sonland\u0131r\u0131n:<\/strong> \u0130\u015flem tamamland\u0131\u011f\u0131nda endSession()<\/code> \u00e7a\u011f\u0131rarak kaynaklar\u0131 serbest b\u0131rak\u0131n.<\/p>\nconst { MongoClient } = require('mongodb');\n\nasync function runTransaction() {\n const uri = \"mongodb:\/\/localhost:27017\"; \/\/ veya replika set URI'niz\n const client = new MongoClient(uri);\n\n try {\n await client.connect();\n const db = client.db(\"banka\");\n\n \/\/ 1. Bir client session ba\u015flat\u0131n\n const session = client.startSession();\n\n \/\/ 2. \u0130\u015flem i\u00e7in readConcern ve writeConcern ayarlar\u0131n\u0131 belirleyin\n const transactionOptions = {\n readConcern: { level: 'snapshot' },\n writeConcern: { w: 'majority' }\n };\n\n try {\n \/\/ 3. \u0130\u015flemi ba\u015flat\u0131n\n session.startTransaction(transactionOptions);\n\n const accountsCollection = db.collection(\"accounts\");\n const transactionsCollection = db.collection(\"transactions\");\n\n const fromAccountId = \"user123\";\n const toAccountId = \"user456\";\n const amount = 100;\n\n \/\/ 4. Operasyonlar\u0131 ger\u00e7ekle\u015ftirin (session'\u0131 kullanarak)\n \/\/ G\u00f6nderen hesab\u0131n bakiyesini azalt\n const updateSender = await accountsCollection.updateOne(\n { _id: fromAccountId, balance: { $gte: amount } },\n { $inc: { balance: -amount } },\n { session } \/\/ \u00d6NEML\u0130: Operasyonlara session'\u0131 ge\u00e7in\n );\n\n if (updateSender.matchedCount === 0) {\n \/\/ Yetersiz bakiye veya hesap bulunamad\u0131\n throw new Error(\"G\u00f6nderen hesap bulunamad\u0131 veya yetersiz bakiye.\");\n }\n\n \/\/ Al\u0131c\u0131 hesab\u0131n bakiyesini art\u0131r\n const updateReceiver = await accountsCollection.updateOne(\n { _id: toAccountId },\n { $inc: { balance: amount } },\n { session }\n );\n\n if (updateReceiver.matchedCount === 0) {\n throw new Error(\"Al\u0131c\u0131 hesap bulunamad\u0131.\");\n }\n\n \/\/ \u0130\u015flem kayd\u0131n\u0131 olu\u015ftur\n await transactionsCollection.insertOne(\n {\n from: fromAccountId,\n to: toAccountId,\n amount: amount,\n date: new Date(),\n status: \"completed\"\n },\n { session }\n );\n\n \/\/ 5. \u0130\u015flemi commit edin\n await session.commitTransaction();\n console.log(\"Para transferi ve i\u015flem kayd\u0131 ba\u015far\u0131yla tamamland\u0131.\");\n\n } catch (error) {\n \/\/ 6. Hata durumunda i\u015flemi iptal edin\n await session.abortTransaction();\n console.error(\"\u0130\u015flem ba\u015far\u0131s\u0131z oldu:\", error.message);\n } finally {\n \/\/ 7. Session'\u0131 sonland\u0131r\u0131n\n await session.endSession();\n }\n\n } finally {\n await client.close();\n }\n}\n\nrunTransaction().catch(console.dir);<\/code><\/pre>\nHata Y\u00f6netimi ve Yeniden Deneme Mant\u0131\u011f\u0131<\/h4>\n
TransientTransactionError<\/code> ve UnknownTransactionCommitResult<\/code>) sa\u011flar. Bu hatalar al\u0131nd\u0131\u011f\u0131nda, i\u015flem mant\u0131\u011f\u0131n\u0131 yeniden denemek genellikle g\u00fcvenli ve \u00f6nerilen bir yakla\u015f\u0131md\u0131r.<\/p>\nwithTransaction<\/code> gibi yard\u0131mc\u0131 fonksiyonlar sunar. E\u011fer s\u00fcr\u00fcc\u00fcn\u00fczde bu yoksa, manuel bir d\u00f6ng\u00fc ile yeniden deneme mant\u0131\u011f\u0131 uygulaman\u0131z gerekir.<\/p>\n\/\/ Node.js s\u00fcr\u00fcc\u00fcs\u00fcndeki withTransaction \u00f6rne\u011fi (varsa)\nconst { MongoClient } = require('mongodb');\n\nasync function runTransactionWithRetry() {\n const uri = \"mongodb:\/\/localhost:27017\";\n const client = new MongoClient(uri);\n\n try {\n await client.connect();\n const db = client.db(\"banka\");\n const session = client.startSession();\n\n const transactionOptions = {\n readConcern: { level: 'snapshot' },\n writeConcern: { w: 'majority' }\n };\n\n try {\n await session.withTransaction(async () => {\n const accountsCollection = db.collection(\"accounts\");\n const transactionsCollection = db.collection(\"transactions\");\n\n const fromAccountId = \"user123\";\n const toAccountId = \"user456\";\n const amount = 100;\n\n const updateSender = await accountsCollection.updateOne(\n { _id: fromAccountId, balance: { $gte: amount } },\n { $inc: { balance: -amount } },\n { session }\n );\n\n if (updateSender.matchedCount === 0) {\n throw new Error(\"G\u00f6nderen hesap bulunamad\u0131 veya yetersiz bakiye.\");\n }\n\n const updateReceiver = await accountsCollection.updateOne(\n { _id: toAccountId },\n { $inc: { balance: amount } },\n { session }\n );\n\n if (updateReceiver.matchedCount === 0) {\n throw new Error(\"Al\u0131c\u0131 hesap bulunamad\u0131.\");\n }\n\n await transactionsCollection.insertOne(\n {\n from: fromAccountId,\n to: toAccountId,\n amount: amount,\n date: new Date(),\n status: \"completed\"\n },\n { session }\n );\n console.log(\"Para transferi ve i\u015flem kayd\u0131 ba\u015far\u0131yla tamamland\u0131.\");\n }, transactionOptions);\n } catch (error) {\n console.error(\"\u0130\u015flem ba\u015far\u0131s\u0131z oldu:\", error.message);\n \/\/ withTransaction, ge\u00e7ici hatalar i\u00e7in otomatik yeniden deneme yapar.\n \/\/ Di\u011fer hatalar\u0131 burada ele alabiliriz.\n } finally {\n await session.endSession();\n }\n } finally {\n await client.close();\n }\n}\n\n\/\/ runTransactionWithRetry().catch(console.dir); \/\/ E\u011fer withTransaction destekleniyorsa<\/code><\/pre>\nwithTransaction<\/code> gibi bir yard\u0131mc\u0131 fonksiyon yoksa, manuel olarak bir yeniden deneme d\u00f6ng\u00fcs\u00fc olu\u015fturman\u0131z gerekir:<\/p>\n\/\/ Manuel yeniden deneme d\u00f6ng\u00fcs\u00fc \u00f6rne\u011fi\nasync function commitWithRetry(session) {\n try {\n await session.commitTransaction();\n console.log(\"\u0130\u015flem ba\u015far\u0131yla commit edildi.\");\n } catch (error) {\n if (error.has<\/error>ErrorLabel('UnknownTransactionCommitResult')) {\n \/\/ Commit durumu bilinmiyorsa, yeniden dene\n console.log(\"UnknownTransactionCommitResult hatas\u0131, yeniden deneme...\");\n await commitWithRetry(session);\n } else {\n throw error;\n }\n }\n}\n\nasync function runTransactionWithManualRetry() {\n const uri = \"mongodb:\/\/localhost:27017\";\n const client = new MongoClient(uri);\n\n try {\n await client.connect();\n const db = client.db(\"banka\");\n\n let transactionAborted = false;\n while (true) { \/\/ Yeniden deneme d\u00f6ng\u00fcs\u00fc\n const session = client.startSession();\n const transactionOptions = {\n readConcern: { level: 'snapshot' },\n writeConcern: { w: 'majority' }\n };\n\n try {\n session.startTransaction(transactionOptions);\n\n const accountsCollection = db.collection(\"accounts\");\n const transactionsCollection = db.collection(\"transactions\");\n\n const fromAccountId = \"user123\";\n const toAccountId = \"user456\";\n const amount = 100;\n\n const updateSender = await accountsCollection.updateOne(\n { _id: fromAccountId, balance: { $gte: amount } },\n { $inc: { balance: -amount } },\n { session }\n );\n\n if (updateSender.matchedCount === 0) {\n throw new Error(\"G\u00f6nderen hesap bulunamad\u0131 veya yetersiz bakiye.\");\n }\n\n const updateReceiver = await accountsCollection.updateOne(\n { _id: toAccountId },\n { $inc: { balance: amount } },\n { session }\n );\n\n if (updateReceiver.matchedCount === 0) {\n throw new Error(\"Al\u0131c\u0131 hesap bulunamad\u0131.\");\n }\n\n await transactionsCollection.insertOne(\n {\n from: fromAccountId,\n to: toAccountId,\n amount: amount,\n date: new Date(),\n status: \"completed\"\n },\n { session }\n );\n\n await commitWithRetry(session); \/\/ Commit ve yeniden deneme\n console.log(\"\u0130\u015flem ba\u015far\u0131yla tamamland\u0131.\");\n break; \/\/ Ba\u015far\u0131l\u0131, d\u00f6ng\u00fcden \u00e7\u0131k\n } catch (error) {\n await session.abortTransaction();\n console.error(\"\u0130\u015flem ba\u015far\u0131s\u0131z oldu:\", error.message);\n if (error.has<\/error>ErrorLabel('TransientTransactionError')) {\n console.log(\"TransientTransactionError hatas\u0131, yeniden deneme...\");\n transactionAborted = true;\n \/\/ session.endSession() burada \u00e7a\u011fr\u0131lmaz, yeni bir session i\u00e7in d\u00f6ng\u00fc devam eder\n } else {\n throw error; \/\/ Kal\u0131c\u0131 hata, d\u00f6ng\u00fcden \u00e7\u0131k\n }\n } finally {\n if (!transactionAborted) { \/\/ E\u011fer bir TransientTransactionError olu\u015fmad\u0131ysa session'\u0131 sonland\u0131r\n await session.endSession();\n }\n }\n }\n } finally {\n await client.close();\n }\n}\n\n\/\/ runTransactionWithManualRetry().catch(console.dir);<\/code><\/pre>\nTransientTransactionError<\/code> ve UnknownTransactionCommitResult<\/code> hatalar\u0131n\u0131 yakalay\u0131p i\u015flemi yeniden deneme mant\u0131\u011f\u0131n\u0131 g\u00f6sterir. Ger\u00e7ek d\u00fcnya uygulamalar\u0131nda, yeniden deneme say\u0131s\u0131 s\u0131n\u0131r\u0131 veya gecikme (backoff) stratejileri gibi ek mekanizmalar da eklemek isteyebilirsiniz.<\/p>\nRead Concern ve Write Concern Ayarlar\u0131<\/h4>\n
readConcern<\/code> ve writeConcern<\/code> ayarlar\u0131 belirtilebilir. Bu ayarlar, i\u015flem i\u00e7indeki t\u00fcm okuma ve yazma operasyonlar\u0131 i\u00e7in ge\u00e7erli olur.
\n* readConcern: { level: 'snapshot' }<\/code><\/strong>: \u0130\u015flemdeki okuma operasyonlar\u0131n\u0131n, i\u015flemin ba\u015flad\u0131\u011f\u0131 andaki verinin bir anl\u0131k g\u00f6r\u00fcnt\u00fcs\u00fcn\u00fc g\u00f6rmesini sa\u011flar. Bu, i\u015flem boyunca tutarl\u0131 okumalar i\u00e7in kritik \u00f6neme sahiptir.
\n* writeConcern: { w: 'majority' }<\/code><\/strong>: \u0130\u015flemin commit edilmesi i\u00e7in replika setindeki \u00e7o\u011fu d\u00fc\u011f\u00fcme yazma i\u015fleminin onaylanmas\u0131n\u0131 bekler. Bu, veri kal\u0131c\u0131l\u0131\u011f\u0131 ve dayan\u0131kl\u0131l\u0131\u011f\u0131 i\u00e7in en g\u00fcvenli ayard\u0131r.<\/p>\nlinearizable<\/code> read concern, en g\u00fc\u00e7l\u00fc tutarl\u0131l\u0131\u011f\u0131 sa\u011flar; okumalar\u0131n, t\u00fcm \u00f6nceki ba\u015far\u0131l\u0131 yazma i\u015flemlerinin etkilerini g\u00f6rmesini garanti eder. Ancak bu, performans\u0131 olumsuz etkileyebilir ve genellikle snapshot<\/code> yeterlidir.<\/p>\nMongoDB \u0130\u015flemlerinde Dikkat Edilmesi Gerekenler ve S\u0131n\u0131rlamalar<\/h3>\n
Performans Etkisi<\/h4>\n
\n* Kilitlenme:<\/strong> \u0130\u015flemler, eri\u015fti\u011fi belgeler \u00fczerinde kilitler tutar. Uzun s\u00fcreli veya \u00e7ok say\u0131da belgeyi etkileyen i\u015flemler, di\u011fer operasyonlar\u0131 engelleyerek performans\u0131 d\u00fc\u015f\u00fcrebilir.
\n* A\u011f Trafi\u011fi:<\/strong> \u0130\u015flemler, session ve commit\/abort mesajlar\u0131 nedeniyle ek a\u011f trafi\u011fi olu\u015fturur.
\n* Replikasyon:<\/strong> \u0130\u015flemlerin commit edilmesi, writeConcern<\/code> ayarlar\u0131na g\u00f6re replika set \u00fcyelerine yay\u0131lmas\u0131n\u0131 gerektirir, bu da gecikmeye neden olabilir.<\/p>\n\u0130\u015flem S\u00fcresi S\u0131n\u0131rlar\u0131<\/h4>\n
transactionLifetimeLimitSeconds<\/code>, varsay\u0131lan 60 saniye). Bir i\u015flem bu s\u00fcreyi a\u015farsa, otomatik olarak iptal edilir. Bu ayar mongod<\/code> yap\u0131land\u0131rmas\u0131nda de\u011fi\u015ftirilebilir, ancak genellikle i\u015flemleri daha k\u0131sa tutmak daha iyi bir yakla\u015f\u0131md\u0131r.<\/p>\nKoleksiyonlar Aras\u0131 ve Sharded Cluster \u0130\u015flemleri<\/h4>\n
Desteklenmeyen Operasyonlar<\/h4>\n
\n* Veritaban\u0131 ve Koleksiyon Y\u00f6netimi:<\/strong>
\n * createCollection<\/code> (MongoDB 4.2+ ile createCollection<\/code> bir i\u015flem i\u00e7inde desteklenir, ancak varsay\u0131lan olarak db.createCollection()<\/code> i\u015flemi d\u0131\u015f\u0131nda \u00e7al\u0131\u015f\u0131r. Session ile \u00e7a\u011fr\u0131lmas\u0131 gerekir.)
\n * createIndex<\/code>, listCollections<\/code>, listDatabases<\/code>
\n * dropDatabase<\/code> (MongoDB 4.2+ ile dropCollection<\/code> desteklenir, ancak dropDatabase<\/code> hala desteklenmez.)
\n* Sistem Komutlar\u0131:<\/strong>
\n * killCursors<\/code>, authenticate<\/code>, getParameter<\/code>, setParameter<\/code> gibi \u00e7o\u011fu y\u00f6netim komutu.
\n* count<\/code> Operasyonu:<\/strong> count<\/code> metodu i\u015flem i\u00e7inde kullan\u0131lamaz. Bunun yerine aggregation pipeline’daki $count<\/code> operat\u00f6r\u00fc veya estimatedDocumentCount<\/code> metodu kullan\u0131labilir.
\n* findAndModify<\/code> Operasyonu:<\/strong> findAndModify<\/code> i\u015flemi i\u00e7inde desteklenmez. Bunun yerine findOne<\/code> ile okuma ve ard\u0131ndan updateOne<\/code> veya deleteOne<\/code> ile g\u00fcncelleme\/silme i\u015flemleri kullan\u0131lmal\u0131d\u0131r.<\/p>\nShard Key De\u011fi\u015fiklikleri<\/h4>\n
Tek Belge Atomikli\u011fi<\/h4>\n
Driver Deste\u011fi<\/h4>\n
En \u0130yi Uygulamalar (Best Practices)<\/h3>\n
\n* Do\u011fru Read\/Write Concern Ayarlar\u0131n\u0131 Kullan\u0131n:<\/strong> \u00c7o\u011fu senaryo i\u00e7in writeConcern: { w: 'majority' }<\/code> ve readConcern: { level: 'snapshot' }<\/code> g\u00fcvenilir ve dengeli bir performans sunar. linearizable<\/code> read concern daha g\u00fc\u00e7l\u00fc tutarl\u0131l\u0131k sa\u011flasa da, performans maliyeti nedeniyle yaln\u0131zca kesinlikle gerekti\u011finde kullan\u0131lmal\u0131d\u0131r.
\n* Hata Y\u00f6netimi ve Yeniden Deneme Mant\u0131\u011f\u0131 Uygulay\u0131n:<\/strong> \u00d6zellikle da\u011f\u0131t\u0131k sistemlerde TransientTransactionError<\/code> ve UnknownTransactionCommitResult<\/code> gibi ge\u00e7ici hatalar ka\u00e7\u0131n\u0131lmazd\u0131r. Bu hatalar i\u00e7in sa\u011flam bir yeniden deneme mekanizmas\u0131 uygulamak, uygulaman\u0131z\u0131n dayan\u0131kl\u0131l\u0131\u011f\u0131n\u0131 art\u0131r\u0131r. S\u00fcr\u00fcc\u00fcn\u00fcn withTransaction<\/code> gibi yard\u0131mc\u0131 fonksiyonlar\u0131n\u0131 kullanmak i\u015fi kolayla\u015ft\u0131rabilir.
\n* \u0130\u015flemdeki Veri Miktar\u0131n\u0131 S\u0131n\u0131rlay\u0131n:<\/strong> \u0130\u015flemler i\u00e7inde b\u00fcy\u00fck veri k\u00fcmeleri \u00fczerinde operasyon yapmaktan ka\u00e7\u0131n\u0131n. Bu, performans\u0131 d\u00fc\u015f\u00fcrebilir ve kilitlenme s\u00fcresini art\u0131rabilir.
\n* Gereksiz \u0130\u015flemlerden Ka\u00e7\u0131n\u0131n:<\/strong> E\u011fer bir operasyon tek bir belgeyi etkiliyorsa, MongoDB’nin tek belge atomikli\u011finden faydalan\u0131n ve ek bir i\u015fleme gerek duymay\u0131n. \u0130\u015flemleri yaln\u0131zca birden fazla belge veya koleksiyon aras\u0131nda atomiklik gerekti\u011finde kullan\u0131n.
\n* Monit\u00f6rizasyon:<\/strong> \u0130\u015flemlerin performans\u0131n\u0131 ve durumunu d\u00fczenli olarak izleyin. Uzun s\u00fcreli veya s\u0131k ba\u015far\u0131s\u0131z olan i\u015flemler, altta yatan sorunlara i\u015faret edebilir. MongoDB Atlas gibi ara\u00e7lar veya db.currentOp()<\/code> gibi komutlar, aktif i\u015flemleri izlemek i\u00e7in kullan\u0131labilir.
\n* Uygulama Mant\u0131\u011f\u0131n\u0131 Basit Tutun:<\/strong> \u0130\u015flem i\u00e7indeki kodun m\u00fcmk\u00fcn oldu\u011funca basit ve odaklanm\u0131\u015f olmas\u0131n\u0131 sa\u011flay\u0131n. A\u011f \u00e7a\u011fr\u0131lar\u0131, uzun s\u00fcreli hesaplamalar veya harici servislerle etkile\u015fim gibi yava\u015f operasyonlardan ka\u00e7\u0131n\u0131n. Bu t\u00fcr operasyonlar i\u015flemin s\u00fcresini uzat\u0131r ve kilitlenmeleri art\u0131r\u0131r.<\/p>\nSonu\u00e7<\/h3>\n