Lapho enza ingqalasizinda yenethiwekhi, umuntu ngokuvamile ucabangela ikhompyutha yendawo noma i-cloud computing. Kodwa lezi zinketho ezimbili nezinhlanganisela zazo zimbalwa. Isibonelo, yini okufanele uyenze uma ungakwazi ukwenqaba i-cloud computing, kodwa awukho umkhawulokudonsa owanele noma ithrafikhi ibiza kakhulu?
Engeza okumaphakathi okuzokwenza ingxenye yezibalo emaphethelweni enethiwekhi yendawo noma inqubo yokukhiqiza. Lo mqondo onqenqemeni ubizwa nge-Edge Computing. Umqondo uhambisana nemodeli yamanje yokusetshenziswa kwedatha yefu, futhi kulesi sihloko sizobheka ihadiwe edingekayo kanye nemisebenzi eyisibonelo yayo.
Amazinga ekhompiyutha asemaphethelweni
Ake sithi unenqwaba yezinzwa ezifakiwe ekhaya: i-thermometer, i-hygrometer, inzwa yokukhanya, inzwa yokuvuza, njalonjalo. Isilawuli esinengqondo sicubungula ulwazi olutholwe kubo, sisebenzisa izimo ezizenzakalelayo, sikhipha i-telemetry ecutshunguliwe kusevisi yefu futhi sithola izimo ezibuyekeziwe zokuzenzakalela kanye ne-firmware entsha kuyo. Ngakho, i-computing yendawo yenziwa ngokuqondile esizeni, kodwa okokusebenza kulawulwa ku-node ehlanganisa amadivaysi amaningi anjalo.
Lesi isibonelo sesistimu yekhompyutha enqenqemeni elula kakhulu, kodwa isivele ikhombisa wonke amazinga amathathu e-edge computing:
- Amadivayisi e-IoT: akhiqiza "idatha eluhlaza" futhi ayidlulise ngezivumelwano ezahlukahlukene.
- Ama-Edge nodes: Cubungula idatha eduze nemithombo yolwazi futhi usebenze njengezitolo zedatha yesikhashana.
- Amasevisi wamafu: nikeza imisebenzi yokuphatha yawo womabili amadivayisi we-peripheral kanye ne-IoT, yenza ukugcinwa kwedatha yesikhathi eside nokuhlaziya. Ngaphezu kwalokho, basekela ukuhlanganiswa nezinye izinhlelo zezinkampani.
Umqondo we-Edge computing ngokwawo uyingxenye ye-ecosystem enkulu eyenza inqubo yezobuchwepheshe ibe ngcono. Kuhlanganisa kokubili i-hardware (amaseva wokubeka nama-edge), kanye nenethiwekhi nezingxenye zesofthiwe (isibonelo, iplatifomu ukuthuthukisa i-AI algorithms). Njengoba amabhodlela angavela ngesikhathi sokudala, ukudluliswa nokucutshungulwa kwedatha enkulu futhi kukhawule ukusebenza kwalo lonke uhlelo, lezi zingxenye kufanele zihambisane nezinye.
Izici zamaseva onqenqema
Ezingeni le-node enqenqemeni, i-Edge Computing isebenzisa amaseva asemaphethelweni abekwe ngqo lapho kukhiqizwa khona ulwazi. Ngokuvamile lezi yizakhiwo zokukhiqiza noma zobuchwepheshe lapho kungenakwenzeka ukufaka i-server rack nokuqinisekisa ukuhlanzeka. Ngakho-ke, amaseva asemaphethelweni agcinwa ezindaweni ezihlangene, ezinothuli-nanomswakama ezinobubanzi obushushu obunwetshiwe; azikwazi ukufakwa endaweni yokubeka. Yebo, iseva enjalo ingalenga kalula kumahange ethayipu anezinhlangothi ezimbili endaweni ethile ngaphansi kwezitebhisi noma egumbini lezinsiza.
Njengoba amaseva onqenqema efakwe ngaphandle kwezikhungo zedatha evikelekile, anezidingo zokuphepha ezingokwenyama eziphezulu. Iziqukathi zokuzivikela zihlinzekelwe bona:
Ezingeni lokucubungula idatha, amaseva asemaphethelweni ahlinzeka ngokubethela kwediski kanye nokuqalisa okuphephile. Ukubethela ngokwako kudla u-2-3% wamandla ekhompyutha, kodwa amaseva asemaphethelweni avamise ukusebenzisa amaphrosesa e-Xeon D anemojula yokusheshisa eyakhelwe ngaphakathi ye-AES, enciphisa ukulahleka kwamandla.
Lapho Usebenzisa Amaseva E-Edge
Nge-Edge Computing, isikhungo sedatha sithola ukuze kucutshungulwe kuphela leyo datha engenakwenzeka noma engenangqondo ukucutshungulwa nganoma iyiphi enye indlela. Ngakho, amaseva asemaphethelweni asetshenziswa uma kudingeka:
- Indlela eguquguqukayo yokuphepha, njengoba endabeni ye-Edge Computing ungalungisa ukudluliswa kolwazi oselucutshungulwe futhi olulungiselelwe esikhungweni sedatha esimaphakathi;
- Ukuvikelwa ekulahlekelweni kolwazi, ngoba uma ukuxhumana nesikhungo kulahleka, ama-node wendawo azoqoqa ulwazi;
- Ukonga kuthrafikhi kufinyelelwa ngokucubungula inqwaba yolwazi esizeni.
I-Edge computing ukuze wonge ithrafikhi
Inkampani yaseDenmark iMaersk, engomunye wabahamba phambili kwezokuthutha olwandle emhlabeni, inqume ukunciphisa ukusetshenziswa kukaphethiloli emikhunjini yayo kanye nokunciphisa ukukhishwa kwezinto ezingcolile emoyeni.
Ubuchwepheshe busetshenziswe ukuxazulula le nkinga , izinzwa ezinjinini kanye nezingxenye eziyinhloko zomkhumbi, kanye neseva yendawo ye-BullSequana Edge yekhompyutha esesayithi.
Ngenxa yezinzwa, uhlelo lwe-EcoMain Suite luhlala luqapha isimo sezingxenye ezibucayi zomkhumbi nokuchezuka kwazo esimisweni esibalwe ngaphambili. Lokhu kukuvumela ukuthi uhlole ngokushesha iphutha futhi ulenze libe lelasendaweni phansi endaweni yenkinga. Njengoba i-telemetry ihlale idluliselwa "esikhungweni", uchwepheshe wesevisi angakwazi ukuhlaziya ekude futhi enze izincomo kubasebenzi abasebhodini. Futhi umbuzo oyinhloko lapha ukuthi ingakanani idatha nokuthi iyiphi ivolumu okufanele idluliselwe esikhungweni sedatha esimaphakathi.
Njengoba ukuxhuma i-inthanethi enezintambo ezishibhile emkhunjini weziqukathi zasolwandle kuyinkinga kakhulu, ukudlulisa inani elikhulu ledatha eluhlaza kuseva emaphakathi kubiza kakhulu. Kuseva ye-BullSequana S200 emaphakathi, imodeli enengqondo iyonke yomkhumbi ibalwa, futhi ukucubungula idatha nokulawula okuqondile kudluliselwa kuseva yendawo. Ngenxa yalokho, ukuqaliswa kwalolu hlelo kuzikhokhele ezinyangeni ezintathu.
I-Edge computing ukuze ulondoloze izinsiza
Esinye isibonelo se-edge computing ukuhlaziya kwevidiyo. Ngakho-ke, kumkhiqizi wemishini yamagesi ezobuchwepheshe i-Air Liquide, omunye wemisebenzi yendawo yomjikelezo wokukhiqiza ukulawula ikhwalithi yokudweba kwamasilinda egesi. Kwenziwa mathupha futhi kwathatha cishe imizuzu eyi-7 isilinda.
Ukuze kusheshiswe le nqubo, umuntu wathathelwa indawo ibhulokhi lamakhamera wevidiyo ayi-7 anencazelo ephezulu. Amakhamera aqopha ibhaluni ezinhlangothini ezimbalwa, akhiqize cishe u-1 GB wevidiyo ngomzuzu. Ividiyo ithunyelwa kuseva ye-BullSequana Edge nge-Nvidia T4 ebhodini, lapho inethiwekhi ye-neural eqeqeshelwe ukucinga okungalungile ihlaziya ukusakaza ku-inthanethi. Ngenxa yalokho, isikhathi sokuhlola esijwayelekile sincishisiwe ukusuka emaminithini ambalwa ukuya kumasekhondi ambalwa.
I-Edge computing in analytics
Ukugibela e-Disneyland akuyona nje into ejabulisayo, kodwa futhi izinto zobuchwepheshe eziyinkimbinkimbi. Ngakho-ke, ku-"Roller Coaster" kufakwe izinzwa ezihlukene ezingaba ngu-800. Bahlala bethumela idatha emayelana nokusebenza kokukhanga kuseva, futhi iseva yendawo icubungula le datha, ibala amathuba okukhanga ehlulekayo, futhi ikubonisa lokhu esikhungweni sedatha esimaphakathi.
Ngokusekelwe kule datha, amathuba okuhluleka kobuchwepheshe ayanqunywa futhi ukulungiswa kokuvimbela kuyaqaliswa. Ukukhanga kuyaqhubeka nokusebenza kuze kube sekupheleni kosuku lokusebenza, futhi okwamanje umyalelo wokulungisa usukhishiwe, futhi abasebenzi balungisa ngokushesha ukukhanga ebusuku.
I-BullSequana Edge
Amaseva e-BullSequana Edge ayingxenye yengqalasizinda enkulu yokusebenza βngedatha enkuluβ; asehloliwe kakade namapulatifomu e-Microsoft Azure kanye ne-Siemens MindSphere, i-VMware WSX futhi anezitifiketi ze-NVidia NGC/EGX. Lawa maseva aklanyelwe ngokuqondile i-edge computing futhi atholakala ku-U2 form factor chassis ku-rack evamile, i-DIN ujantshi, izinketho zodonga kanye ne-tower mount.
I-BullSequana Edge yakhelwe ebhodini lobunikazi lobunikazi kanye nephrosesa ye-Intel Xeon D-2187NT. Basekela ukufakwa kokufika ku-512 GB we-RAM, ama-SSD angu-2 we-960 GB noma ama-HDD angu-2 we-8 noma i-14 TB. Bangakwazi futhi ukufaka i-2 Nvidia T4 16 GB GPUs yokucubungula ividiyo; Amamojula we-Wi-fi, i-LoRaWAN ne-4G; kufika kumamojula angu-2 10-Gigabit SFP. Amaseva ngokwawo asenayo kakade inzwa yokuvula isivalo efakiwe, exhunywe ku-BMC elawula imojula ye-IPMI. Ingalungiselelwa ukuvala amandla ngokuzenzakalela uma inzwa icushiwe.
Imininingwane egcwele yobuchwepheshe yamaseva e-BullSequana Edge ingatholakala kokuthi . Uma unentshisekelo ngemininingwane, sizokujabulela ukuphendula imibuzo yethu kumazwana.
Source: www.habr.com