I-iFood, inkundla enkulu yokulethwa kokudla e-Latin America, ichaza kabusha ukuthi isebenzisa kanjani idatha ukuze ishayele emisha nokusebenza kahle ngosizo lwe-Databricks, idatha kanye nenkampani ye-AI. Le nkampani yamukele inkundla yezobunhloli bedatha yeDatabricks kanye nohlaka lwayo lwe-DLT, olwakha futhi luhlole amapayipi edatha athembekile, ukuze kwenziwe kube manje ukwakhiwa kwayo kwedatha.
Lo mzamo uholele ochungechungeni lwezinzuzo zokusebenza, okuhlanganisa ukuncishiswa kwezindleko zokucubungula nokugcina ngo-67%, umzamo ongaphansi ngo-70% ekulungisweni kwamapayipi, kanye nokwehliswa ngo-30% esikhathini sokuthuthukiswa kwekhodi.
"Amapayipi manje asebenza ngokungenasici, aletha ukusebenza okuthembekile ngisho nangaphansi komsebenzi onzima kakhulu," kusho u-Thiago Julião, uchwepheshe wezokwakha idatha kwa-iFood. "Ukusuka emaphutheni avamile kuye ezinkingeni ezicishe zibe ziro ngokuthuthela ku-DLT akugcini nje ngokuthuthukisa ukusebenza kahle kodwa kukhulule nethimba lethu ukuthi ligxile ezinhlelweni zamasu esikhundleni sokucisha imililo."
Nge-ecosystem eqinile - abasebenzisi abangaphezu kwezigidi ezingu-55, izindawo zokudlela zozakwethu ezingu-350,000, nezishayeli zokulethwa ezisebenzayo ezingu-300,000 - inkundla icubungula cishe imicimbi yedatha yesikhathi sangempela cishe eyizigidi eziyizinkulungwane ezingu-10 nsuku zonke, kusukela kuzinhlelo zayo zokusebenza, izinhlelo zokusebenza, nezinkundla zozakwethu.
"I-Databricks iguqule ngokuphelele ukusebenza kwethu kwedatha. Nge-DLT, sehlise kakhulu amaphutha futhi sasusa ukugxila kwethimba lethu kuzinhlelo zamasu esikhundleni sezinkinga zokusebenza," kusho u-Julião.
Kusukela endaweni ehlukene kuya esakhiweni esihlanganisiwe
Ngaphambi kokuqaliswa, inkampani ibhekane nezinselelo ngezinqubo zezakhiwo ezihlukene kanye nezandla, ezivame ukuholela emaphutheni futhi zivimbele ukukhula. Ukuthuthela kunkundla kuhlanganise izinkulungwane zamathebula kwaba ngu-100 nje, kuthuthukisa ukubusa nokuthuthukisa ikhwalithi yedatha.
"Izixazululo zethu ziklanyelwe ukushayela ukukhula kwebhizinisi kumakhasimende ethu ngokunciphisa izindleko nokwandisa ukusebenza kahle. Ngedatha ekhiqizwayo kanye nesimanje nge-DLT, i-iFood inokufinyelela emininingwaneni yamasu efaka isandla ekuthuthukiseni ulwazi lwamakhasimende lapho efaka ama-oda," kuchaza uMarcos Grilanda, iPhini Likamongameli kanye noMphathi-Jikelele weLatin America.
Namuhla, ngendlela ehlelekile nenezingqimba, inkundla iqinisekisa ukungeniswa kwedatha nge-latency yezinga lesibili, ukuqinisekiswa okuzenzakalelayo, nokufinyelela okulula kwezindawo zebhizinisi. Lokhu kuvumela ukuhlaziya kwesikhathi sangempela, ukuhlola kwe-A/B, nezinqumo eziqhutshwa idatha ezithinta ngokuqondile ulwazi lwamakhasimende nokusebenza kahle.
"Izinzuzo zokusebenza ohambweni lomsebenzisi zishintshile. Manje sesinokulawula okukhulu, isivinini, nokuthembeka kokuthi idatha isetshenziswa kanjani kuyo yonke inkampani," kusho u-Maristela Albuquerque, Umphathi Wedatha kwa-iFood.
Izinyathelo ezilandelayo: ukuphepha, ukukala, kanye nokuqamba okusha okuqhubekayo.
Ngesisekelo esiqinile sobuchwepheshe, i-iFood ihlela ukwandisa ukusetshenziswa kwenkundla ngezici ezifana ne-Databricks Asset Bundles (DABs), ukusebenza okungenasiphakeli, nokuvala amakholomu azwelayo, okuthuthuke ngokwengeziwe ukubusa nokuphepha kwedatha.
"Lolu shintsho luye lwasivumela ukuba siqede ukungasebenzi kahle, sisheshise ukuthuthukiswa kwezixazululo futhi, ngaphezu kwakho konke, sakhe isisekelo esiqinile sokwenza izinto ezintsha eziqhubekayo," kuphetha uGabriel Campos, iNhloko Yedatha kanye ne-AI ku-iFood.
