![Search icon](https://hackernoon.imgix.net/search-new.png?w=19&h=19)
Uma uhlangane naleli khasi ucabanga ukuthi uzoceba ngohlelo oluthile lokuceba ngokushesha, ngiyaxolisa ukukudumaza. Lesi sihloko sizokhuluma ngokuthi unganciphisa kanjani izikweletu zakho zezindleko zamafu ngo-$1 million. Ngokwenza lokho, empeleni uzobe ukhiqize izigidi ezengeziwe zamadola emalini engenayo - ongayisebenzisa ukuthenga isifundo sami se-inthanethi sokuthi ungaceba kanjani nge-AWS ( isixhumanisi sesifundo lapha ).
Izindleko zamafu ngokuvamile azinakwa futhi azibalwa ekuqaleni kwamaphrojekthi ezinkampani. Inhlolovo ye-HashiCorp ka-2021 ithole ukuthi cishe i-40% yezinkampani ezisebenzisa ngokweqile izindleko zamafu ngo-2021 [ 1 ]. Ku-2023, cishe zonke izinkampani (94%) zivumile ukuthi zichitha imali efwini [ 1 ] futhi okungenani i-30% yezindleko zamafu yachithwa [ 2 ]. Ukusetshenziswa kwamafu bekucishe kube ngamabhiliyoni angama-500 ngo-2022 - ngakho-ke sikhuluma ngama- $ 150 ezigidigidi ezimoshwe ngonyaka !!
Akukhona lokhu kuphela ukukhathazeka ngemali engenayo egejile kodwa futhi nemikhuba engemihle yokusimama. AmaRandi ayizigidi eziyizinkulungwane ezingu-150 wamandla amoshiwe!
Lokhu okutholakele kuhilela amabhizinisi amakhulu kanye namancane, kusukela ekuvuthweni kwamafu aphezulu kuya ekuvuthweni kwamafu aphansi. Ibhekisela ku-AWS, kodwa izimiso ezifanayo zingasetshenziswa kunoma yimuphi omunye umhlinzeki wamafu. Ngakho-ke, uma noma iyiphi ingxenye yomsebenzi wakho isefwini, lesi sihloko sesakho.
Ngikhuluma ngombono kanjiniyela wedatha, kodwa ukufunda okufanayo kungasetshenziswa kwezinye izinqubo zobunjiniyela besofthiwe.
Asicwilise phakathi.
Lolu hlobo lwebhili yamafu ngokuvamile lukhawulelwe kumabhizinisi amakhulu kakhulu asebenza emhlabeni jikelele ngezigidi zamakhasimende.
Ukuze sikunikeze umbono, ibhili yefu engu-$1 million ingavela ekucutshungulweni komsebenzi we-Spark ETL ~1.5Tb ngehora 24x7 izinsuku ezingu-365 ngonyaka. Esinye isibonelo kungaba uhlelo lokusebenza oluthola izigidigidi zezicelo ngosuku ezivela ezindaweni eziningi emhlabeni.
Ebhizinisini elikhulu, kunamakhulu ezicelo ngalo sayizi - okuholela ezinkontilekeni zamadola ayizigidi eziyizinkulungwane nabahlinzeki bamafu. Ngokwesibonelo, i-Airbnb yayinesibopho sokusebenzisa u-$1.2 wezigidigidi ezinsizeni zamafu phakathi neminyaka emihlanu ekupheleni kuka-2019 [3 ].
Kwa-Expedia sehlise izindleko ze-ETL yokucubungula idatha ebiza u-$1.1 wesigidi samadola ngonyaka yayisa ku-$100,000 nje ngonyaka ngokusebenzisa izinqubo zokuthuthukisa. Lokho kungukwehliswa kwezindleko ngo-91%.
Akuzona zonke izinkampani ezinezicelo zosayizi omkhulu kangako kodwa cabanga ukusika izindleko zakho zamafu ngo-90% ngohlelo lokusebenza olulodwa noma inkampani yakho yonke.
Hamba uyolanda uhlu lwezinhlelo zakho zokusebenza ezibiza kakhulu futhi uphonsele inselelo imibono yakho yedizayini .
Yonke le mibuzo ibuyela embuzweni obaluleke kakhulu: sizosetshenziswa kanjani isicelo? Lingakanani inani lebhizinisi ukuze libe khona? Uhlelo lokusebenza lusisiza kanjani ukuba sifinyelele umgomo othile?
Yiqiniso, zonke lezi zimpendulo ngokuvamile azicaci kahle ekuqaleni kwephrojekthi; kodwa yingakho ukuklama kufanele kuhlale kuyinqubo ephindaphindayo - evumela izinguquko ukuthi zenzeke ngaphandle komthungo ngangokunokwenzeka. Onjiniyela kufanele bamukele ukuziphendukela kwemvelo kanye noshintsho, baqondanise ukuthuthukiswa kohlelo lokusebenza nomthelela.
Isinyathelo sesibili sihlanganisa ukuhlinzeka uhlelo lokusebenza ngezinsiza ezifanele nokulushuna nengqalasizinda efanele.
Njengonjiniyela, qaphela ukuthi izindleko zamafu zibalwa kanjani. Isibonelo, i-AWS ihlinzeka ngezimo, lapho ungabhida khona intengo yeqoqo - lokhu kubaluleke kakhulu uma unezinhlelo zokusebenza ezibekezelela amaphutha nezivumelana nezimo. Zisebenzise uma ukwazi — I-AWS ifuna ukuncishiswa kwezindleko ngo-90% [ 4 ].
Ezinye izinto ongase ufune ukuzicabangela yilezi:
Kukhona okuncane okungekho okungahambi kahle ekusebenziseni izimo ze-AWS Graviton. I-AWS itshale imali eningi ekudaleni amaphrosesa angabizi kakhulu. Ungathola ukwehliswa okungafika ku-40% ekusetshenzisweni kwamafu ngokuvele ushintshe usuke ku-intel-based processor uye ku-ARM-based processor [ 10 ].
Okuwukuphela kwesexwayiso kulokhu ukuthi uhlelo lwakho lokusebenza ludinga ukusebenzisana namaphrosesa asekelwe ku-ARM iGraviton esebenza kuwo. Uma usebenzelana nesevisi ephethwe njenge-RDS noma i-OpenSearch ngakho-ke akukho nkinga nhlobo ekushintsheni - I-AWS isebenzisana ne-OS eyisisekelo kanye nokuhambisana kohlelo lokusebenza. Uma uzakhela olwakho uhlelo lokusebenza, ungase udinge ukuhlanganisa kabusha iphakheji kuye ngokuthi usebenzisa luphi ulimi - i-Java nezinye izilimi azidingi ushintsho kuyilapho iPython idinga ukunakwa okuthile.
Okokugcina, ungakhohlwa ukuqhubeka nokuqapha izindleko zakho ukuze uthole iziqongo nezimanga ezingalindelekile. Izindleko zosuku 0 lwesicelo sakho zizohluka ezindlekweni zosuku 170. Qiniseka ukuthi ulandelela izinguquko, futhi uyaqonda ukuthi kungani ushintsho lwenzeka: ingabe inqwabelanisa izindleko zesitoreji esingu-s3 noma ikhefu elilodwa nje spike?
Setha izexwayiso ezidingekayo kanye nezincwadi zeziqondiso zokusebenza !
Okubalulekile, sebenzisa omaka bokwabiwa kwezindleko ukuze ulandelele ukusetshenziswa kwemali ngomnyango, iphrojekthi, noma indawo. Gwema ubungozi bokudala ixhaphozi ledatha lapho izindleko zingenakulandeleka noma zidinga uhambo olude phakathi kwamasistimu ahlukene alogi. Kufanele kusheshe futhi kube lula ukubuyela kunoma yiziphi izindleko zohlelo olunikeziwe.
Nomaphi lapho usebenza khona, ukulinganisa ukulethwa kwezici ezintsha nokuthuthukisa zamanje kunzima. Ubani ongakaze acindezelwe ukuthi alethe izici ezintsha ezixakile ngesivinini sokukhanya.
Kodwa-ke, kubalulekile kubo bobabili onjiniyela nabaphathi ukwenza izinqumo ngamabomu nezisebenzayo mayelana namaphrojekthi abo amanje, ukulawula ubungozi namathuba ngempumelelo.