Idatha enkulu yentlawulo enkulu: malunga neBigData kwi-telecom

Ngo-2008, iBigData yayilixesha elitsha kunye nefashoni. Kwi-2019, i-BigData yinto yokuthengiswa, umthombo wenzuzo kunye nesizathu samatyala amatsha.

Ekwindla yokugqibela, urhulumente waseRussia uqalise umthetho oyilwayo wokulawula idatha enkulu. Abantu ngabanye abanakuchongwa kulwazi, kodwa banokukwenza oko ngokucelwa ngabasemagunyeni. Ukusetyenzwa kweBigData kumaqela esithathu kuphela emva kokwaziswa kweRoskomnadzor. Iinkampani ezineedilesi zenethiwekhi ezingaphezu kwe-100 lamawaka ziwela phantsi komthetho. Kwaye, ngokuqinisekileyo, apho ngaphandle kweerejista - kufuneka yenze enye kunye noluhlu lwabaqhubi bedatha. Kwaye ukuba ngaphambi kokuba le Nkcukacha eNkulu ayizange ithathwe ngokungathí sina ngumntu wonke, ngoku kuya kufuneka ithathelwe ingqalelo.

Mna, njengomlawuli wenkampani yomphuhlisi ohlawulayo eqhuba le Nkcukacha inkulu kakhulu, andinakuyihoya idatabase. Ndiza kucinga ngedatha enkulu nge-prism yabaqhubi be-telecom, ngeenkqubo zabo zokuhlawula zihamba ngolwazi malunga namawakawaka ababhalisile abadlula yonke imihla.

Ithiyori

Masiqale, njengakwingxaki yemathematika: okokuqala singqina ukuba idatha yabaqhubi be-telecom inokubizwa ngokuba yiBigDat. Ngokuqhelekileyo, idatha enkulu ibonakala ngeempawu ezintathu zeVVV, nangona kwiitoliko zamahhala inani le "Vs" lifikelele kwisixhenxe.

Umthamo. I-MVNO ye-Rostelecom yodwa ikhonza ngaphezu kwesigidi sababhalisile. Abaqhubi abaphambili ababamba idatha kwi-44 ukuya kwi-78 yezigidi zabantu. I-Traffic ikhula rhoqo ngesibini: kwikota yokuqala ye-2019, ababhalisi sele befikelele kwi-3,3 yezigidigidi ze-GB kwiifowuni eziphathwayo.

Isantya. Akukho mntu unokukuxelela malunga ne-dynamics engcono kunezibalo, ngoko ndiya kuhamba ngeengqikelelo zeCisco. Ngo-2021, i-20% yetrafikhi ye-IP iya kuya kwi-traffic traffic - iya phantse kathathu kwiminyaka emihlanu. Ingxenye yesithathu yoxhumo lweselula luya kuba yi-M2M - ukuphuhliswa kwe-IoT kuya kukhokelela ekunyuseni okuphindwe kathandathu kunxibelelwano. I-Intanethi yeZinto ayiyi kuba nenzuzo kuphela, kodwa kunye nemithombo yobutyebi, ngoko abanye abaqhubi baya kugxila kuyo kuphela. Kwaye abo baphuhlisa i-IoT njengenkonzo eyahlukileyo baya kufumana i-traffic ephindwe kabini.

Iintlobo ngeentlobo. Ukwahluka yingcinga ezimeleyo, kodwa abaqhubi be-telecom bazi ngokwenene phantse yonke into malunga nababhalisi babo. Ukususela kwigama kunye neenkcukacha zepasipoti ukuya kwimodeli yefowuni, ukuthenga, iindawo ezityelelwe kunye nezinto ezinomdla. Ngokomthetho waseYarovaya, iifayile zeendaba zigcinwa kwiinyanga ezintandathu. Ke masiyithathe njenge-axiom yokuba idatha eqokelelweyo iyahluka.

Isoftware kunye nendlela yokusebenza

Ababoneleli ngomnye wabathengi abaphambili beBigData, ke uninzi lweendlela ezinkulu zokuhlalutya idatha ziyasebenza kwishishini le-telecom. Omnye umbuzo kukuba ngubani olungele ukutyalomali ekuphuhliseni i-ML, i-AI, i-Deep Learning, utyalo-mali kumaziko edatha kunye nokumbiwa kwedatha. Umsebenzi opheleleyo kunye nedathabheyisi iquka iziseko zophuhliso kunye neqela, iindleko ezingenakukwazi ukufikelela kuwo wonke umntu. Amashishini asele enendawo yokugcina impahla okanye aphuhlisa indlela yoLawulo lweDatha kufuneka abheje kwiBigData. Kwabo abangekakulungeli utyalo-mali lwexesha elide, ndikucebisa ukuba uyakhe ngokuthe ngcembe i-architecture yesoftware kwaye ufake amacandelo nganye nganye. Ungashiya iimodyuli ezinzima kunye neHadoop okokugqibela. Bambalwa abantu abathenga isisombululo esele senziwe kwiingxaki ezinje ngoMgangatho weDatha kunye neDatha yeMigodi; iinkampani ngokubanzi zilungisa inkqubo ngokweenkcukacha kunye neemfuno zabo - ngokwabo okanye ngoncedo lwabaphuhlisi.

Kodwa ayizizo zonke iintlawulo ezinokuguqulwa ukuze zisebenze neBigData. Okanye kunoko, akuyiyo yonke into kuphela enokuguqulwa. Bambalwa abantu abanokukwenza oku.

Iimpawu ezintathu zokuba inkqubo yokuhlawula inethuba lokuba sisixhobo sokusetyenzwa kwedatha:

• I-Horizontal scalability. Isoftware kufuneka ibe bhetyebhetye - sithetha ngedatha enkulu. Ukunyuka kolwazi kufuneka kuphathwe ngokunyuka ngokulinganayo kwi-hardware kwiqela.
• Ukunyamezela iimpazamo. Iinkqubo ezimandundu ze-prepaid zikholisa ukunyamezela iimpazamo ngokungagqibekanga: ukuhlawula kubekwa kwi-cluster kwii-geolocations ezininzi ukuze ziqinisekisane ngokuzenzekelayo. Kufuneka kwakhona kubekho iikhomputha ezaneleyo kwiqela le-Hadoop xa omnye okanye ngaphezulu behluleka.
• Indawo. Idatha kufuneka igcinwe kwaye iqhutywe kwiseva enye, ngaphandle koko unokuhamba uqhekezeke ekugqithiseni idatha. Enye yeendlela ezithandwayo zemephu-Nciphisa izikimu: iivenkile zeHDFS, iinkqubo zeSpark. Ngokufanelekileyo, isoftware kufuneka idibanise ngokungenamthungo kwiziko ledatha yesiseko kwaye ikwazi ukwenza izinto ezintathu kwenye: ukuqokelela, ukucwangcisa nokuhlalutya ulwazi.

Iqela

Yintoni, njani kwaye yintoni injongo inkqubo iya kuqhuba idatha enkulu igqitywe liqela. Ihlala iquka umntu omnye - isazi sedatha. Nangona, ngokombono wam, iphakheji encinci yabasebenzi kwiDatha enkulu iquka uMphathi weMveliso, iNjineli yeDatha, kunye noMphathi. Owokuqala uyaziqonda iinkonzo, uguqulela ulwimi lobugcisa kulwimi lwabantu kwaye ngokuphambene. Injineli yeDatha izisa iimodeli ebomini isebenzisa iJava/Scala kunye nokulinga ngokuFunda koomatshini. Umphathi ulungelelanisa, abeke iinjongo, kwaye alawule izigaba.

Iingxaki

Kuyinxalenye yeqela leBigData apho iingxaki zidla ngokuvela xa kuqokelela kwaye kusetyenzwa idatha. Inkqubo kufuneka ichaze ukuba iqokelele ntoni kunye nendlela yokuyiqhuba - ukuze uchaze oku, kufuneka uqale uqonde ngokwakho. Kodwa kubaboneleli, izinto azilula kangako. Ndithetha ngeengxaki usebenzisa umzekelo womsebenzi wokunciphisa i-churn yababhalisi - yiloo nto abaqhubi be-telecom abazama ukuyicombulula ngoncedo lweDatha enkulu kwindawo yokuqala.

Ukuzibekela usukelo. Inkcazo yobugcisa ebhalwe kakuhle kunye nokuqonda okuhlukeneyo kwamagama kuye kwaba yintlungu yeenkulungwane zeminyaka kungekhona kuphela kwii-freelancers. Nababhalisi "behla" banokutolikwa ngeendlela ezahlukeneyo - njengabo bangakhange basebenzise iinkonzo zabaqhubi inyanga, iinyanga ezintandathu okanye unyaka. Kwaye ukudala i-MVP esekelwe kwidatha yembali, kufuneka uqonde ukuphindaphinda kweembuyekezo zababhalisi ukusuka kwi-churn - abo bazama abanye abaqhubi okanye bashiya isixeko kwaye basebenzise inani elahlukileyo. Omnye umbuzo obalulekileyo: ixesha elingakanani ngaphambi kokuba umrhumi kulindeleke ukuba ahambe kufuneka umboneleli anqume oku kwaye athathe inyathelo? Iinyanga ezintandathu zisengaphambili kakhulu, iveki ishiywe kakhulu.

Ukutshintshwa kwamagama. Ngokuqhelekileyo, abaqhubi bachonga umxhasi ngenombolo yefowuni, ngoko kunengqiqo ukuba iimpawu kufuneka zilayishwe usebenzisa. Kuthekani ngeakhawunti yakho yobuqu okanye inombolo yesicelo senkonzo? Kuyimfuneko ukugqiba ukuba yeyiphi iyunithi ekufuneka ithathwe njengomthengi ukuze idatha kwinkqubo yomqhubi ingahlukanga. Ukuvavanya ixabiso lomxhasi kwakhona kuyathandabuzeka - nguwuphi umrhumi obaluleke ngakumbi kwinkampani, nguwuphi umsebenzisi ofuna umgudu ongakumbi wokugcina, kwaye ngubani oza "kuwa" kuyo nayiphi na imeko kwaye akukho sizathu sokuchitha izixhobo kubo.

Ukunqongophala kolwazi. Ayingabo bonke abasebenzi ababoneleli abakwaziyo ukuchazela iqela le-BigData ukuba yintoni echaphazela ngokuthe ngqo i-churn yababhalisi kunye nokuba izinto ezinokwenzeka zokuhlawula zibalwa njani. Nokuba bathe igama elinye lazo - i-ARPU - kuvela ukuba ingabalwa ngeendlela ezahlukeneyo: mhlawumbi ngeentlawulo zabathengi ngamaxesha, okanye ngeentlawulo ezizenzekelayo. Kwaye kwinkqubo yomsebenzi, kuphakama eminye imibuzo eyisigidi. Ngaba imodeli igubungela bonke abathengi, lithini ixabiso lokugcina umxhasi, ngaba kukho nayiphi na ingongoma ekucingeni ngezinye iimodeli, kunye nokuba wenze ntoni nabathengi abaye bagcinwa ngempazamo.

Ukumisela iinjongo. Ndiyazazi iintlobo ezintathu zeempazamo zeziphumo ezibangela ukuba abaqhubi bakhathazeke ngesiseko sedatha.

1. Umboneleli utyala imali kwi-BigData, uqhuba iigigabhayithi zolwazi, kodwa ufumana isiphumo ebesinokufunyanwa ngexabiso eliphantsi. Imizobo elula kunye neemodeli, uhlalutyo lwamandulo lusetyenziswa. Iindleko ziphezulu ngokuphindwe kaninzi, kodwa umphumo uyafana.
2. Umsebenzisi ufumana idatha eneenkalo ezininzi njengemveliso, kodwa akayiqondi indlela yokuyisebenzisa. Kukho uhlalutyo - nantsi, iyaqondakala kwaye i-voluminous, kodwa ayisebenzi. Isiphumo sokugqibela, esingenako ukubandakanya injongo "yokucubungula idatha," ayizange icingelwe. Akwanelanga ukucubungula - uhlalutyo kufuneka lube sisiseko sokuhlaziya iinkqubo zoshishino.
3. Imiqobo ekusetyenzisweni kweBigData analytics inokuba yinkqubo yeshishini yakudala kunye nesoftware engafanelekanga kwiinjongo ezintsha. Oku kuthetha ukuba benza iphutha kwinqanaba lokulungiselela - abazange bacinge nge-algorithm yezenzo kunye nezigaba zokwazisa iDatha enkulu emsebenzini.

Kutheni

Ukuthetha ngeziphumo. Ndiza kudlula kwiindlela zokusebenzisa kunye nokwenza imali kwiDatha enkulu esele isetyenziswa ngabaqhubi be-telecom.
Ababoneleli baqikelela kuphela ukuphuma kwababhalisi, kodwa kunye nomthwalo kwizikhululo ezisisiseko.

1. Ulwazi malunga neentshukumo zababhalisi, umsebenzi kunye neenkonzo zefrikhwensi zihlalutywa. Isiphumo: ukuncitshiswa kwenani lokugcwala ngenxa yokwenziwa ngcono kunye nokuphuculwa kweendawo eziyingxaki kwiziseko ezingundoqo.
2. Abaqhubi be-Telecom basebenzisa ulwazi malunga ne-geolocation yababhalisi kunye nokuxinana kwezithuthi xa kuvulwa iindawo zokuthengisa. Ngaloo ndlela, i-BigData analytics sele isetyenziswe yi-MTS kunye neVimpelCom ukucwangcisa indawo yeeofisi ezintsha.
3. Ababoneleli benza imali ngedatha yabo enkulu ngokuyinikezela kumaqela esithathu. Abathengi abaphambili babaqhubi beBigData ziibhanki zorhwebo. Ukusebenzisa i-database, babeka iliso kwimisebenzi ekrokrelayo ye-SIM khadi yombhalisi apho amakhadi adityaniswe khona, kwaye asebenzise amanqaku omngcipheko, ukuqinisekiswa kunye neenkonzo zokubeka iliso. Kwaye ngo-2017, urhulumente waseMoscow wacela intshukumo yokuhamba ngokusekwe kwidatha yeBigData esuka kwiTele2 ukucwangcisa iziseko zophuhliso lobugcisa nezothutho.
4. I-BigData analytics yimigodi yegolide yabathengisi, abanokudala imikhankaso yentengiso yobuqu kumawakawaka amaqela ababhalisile ukuba bakhetha. Iinkampani zeTelecom zidibanisa iiprofayili zentlalo, umdla wabathengi kunye neendlela zokuziphatha zababhalisi, kwaye emva koko zisebenzise iBigData eqokelelweyo ukutsala abathengi abatsha. Kodwa ngokukhuthazwa okukhulu kunye nokucwangciswa kwe-PR, ukuhlawula akusoloko kunomsebenzi owaneleyo: inkqubo kufuneka ngexesha elifanayo ithathele ingqalelo izinto ezininzi ngokuhambelana nolwazi olucacileyo malunga nabaxhasi.

Ngelixa abanye basacinga ukuba i-BigData ibinzana elingenanto, i-Big Four sele isenza imali kuyo. I-MTS ifumana i-ruble ye-14 yeebhiliyoni ukusuka ekuqhutyweni kwedatha enkulu kwiinyanga ezintandathu, kwaye i-Tele2 yonyusa ingeniso evela kwiiprojekthi ngamaxesha amathathu anesiqingatha. I-BigData ijika isuka kwi-trend ibe yinto efunekayo, phantsi kwayo yonke i-telecom operators iya kwakhiwa kwakhona.

umthombo: www.habr.com