Qaphela. transl.: Unjiniyela we-American DevOps uSid Palas, usebenzisa Njengomhlahlandlela wolwazi, ngiqhathanise izindleko zesevisi ye-Managed Kubernetes (ngokulungiselelwa okuhlukile) kusukela kubahlinzeki bamafu abahamba phambili emhlabeni. Inzuzo eyengeziwe yomsebenzi wakhe kwaba ukushicilelwa kwe-Jupyter Notebook ehambisanayo, evumela (ngolwazi oluncane lwePython) ukulungisa izibalo ezenziwe ukuze zivumelane nezidingo zakho.
TL; DR: I-Azure ne-Digital Ocean azikhokhisi ngezinsiza zokubala ezisetshenziselwa indiza yokulawula, okuzenza zibe ukukhetha okuhle kokukhipha amaqoqo amaningi amancane. Ngokusebenzisa inombolo encane yamaqoqo amakhulu, i-GKE ifaneleka kakhulu. Ukwengeza, ungakwazi ukunciphisa kakhulu izindleko ngokusebenzisa ama-spot/preemptive/-low-priority noma "ngokubhalisa" ekusebenziseni isikhathi eside ama-node afanayo (lokhu kusebenza kuzo zonke izinkundla).
Usayizi weqoqo (inani labasebenzi)
Ulwazi jikelele
Isimemezelo se-GKE sokuqala ukushaja amasenti angu-10 ngehora leqoqo ngalinye ngehora leqoqo ngalinye singenze ukuba ngiqale ukuhlaziya intengo yeminikelo emikhulu ephethwe yi-Kubernetes.
Lesi simemezelo sibacasule kakhulu abanye...
Abalingiswa abakhulu be-athikili yilaba:
- I-Google Kubernetes Engine (GCP) β ;
- Isevisi ye-Elastic Kubernetes (AWS) - ;
- Isevisi ye-Azure Kubernetes (Azure) - ;
- Kubernetes ku-Digital Ocean - .
Ukwehlukaniswa Kwezindleko
Izindleko eziphelele zokusebenzisa i-Kubernetes kwenye ngayinye yalezi zinkundla iqukethe izingxenye ezilandelayo:
- Imali yokuphatha iqoqo;
- Ibhalansi yomthwalo (ye-Ingress);
- Izinsiza zekhompyutha (i-vCPU nenkumbulo) yabasebenzi;
- Ukuhamba kwethrafikhi;
- Ukugcina unomphela;
- Ukucubungula idatha ngebhalansi yomthwalo.
Ukwengeza, abahlinzeki bamafu banikeza izaphulelo ezibalulekile uma iklayenti lifuna/lingasebenzisa i-preemptible noma amanodi abaluleke kancane NOMA uzibophezela ukusebenzisa amanodi afanayo iminyaka engu-1-3.
Kuyafaneleka ukugcizelela ukuthi nakuba izindleko ziyisisekelo esihle sokuqhathanisa nokuhlola abahlinzeki bezinsizakalo, ezinye izici kufanele zibhekwe:
- I-Uptime (Isivumelwano Sezinga Lesevisi);
- I-ecosystem yamafu ezungezile;
- Izinguqulo ezitholakalayo zama-K8s;
- Ikhwalithi yemibhalo/ikhithi yamathuluzi.
Nokho, lezi zici zingaphezu kobubanzi balesi sihloko/ucwaningo. IN Izici ezingezona zentengo ze-EKS, i-AKS ne-GKE kuxoxwa ngazo ngokuningiliziwe.
I-Jupyter Notebook
Ukwenza kube lula ukuthola isisombululo esinenzuzo kakhulu, ngithuthukile , usebenzisa plotly + ipywidgets kuwo. Ikuvumela ukuthi uqhathanise okunikezwayo kwabahlinzeki ngosayizi abahlukene beqoqo namasethi wesevisi.
Ungazijwayeza ngenguqulo ebukhoma yencwajana ku-Binder:
Ngazise uma izibalo noma izintengo zangempela zingalungile (lokhu kungenziwa ngodaba noma isicelo sokudonsa ku-GitHub - ).
okutholakele
Maye, maningi kakhulu ama-nuances ukunikeza izincomo eziqondile kunalezo ezifakwe esigabeni se-TL;DR ekuqaleni. Nokho, ezinye iziphetho zisengafinyelelwa:
- Ngokungafani ne-GKE ne-EKS, i-AKS ne-Digital Ocean azikhokhisi ngezinsiza zokulawula isendlalelo. I-AKS ne-DO zinenzuzo ethe xaxa uma i-architecture ihlanganisa amaqoqo amaningi amancane (isibonelo, iqoqo elilodwa wonke unjiniyela noma wonke amaklayenti).
- Izinsiza zekhompiyutha ze-GKE ezingabizi kancane ziyenza ibe nenzuzo enkulu njengoba osayizi beqoqo bekhula*.
- Ukusebenzisa ama-node angabonakali noma ukuhambisana kwe-node yesikhathi eside kunganciphisa izindleko ezingaphezu kuka-50%. Qaphela: I-Digital Ocean ayinikezi lezi zaphulelo.
- Izimali eziphumayo ze-Google ziphezulu, kodwa izindleko zezinsiza zekhompuyutha ziyisici esinqumayo ekubaleni (ngaphandle uma iqoqo lakho likhiqiza inani elibalulekile ledatha ephumayo).
- Ukukhetha izinhlobo zemishini ngokusekelwe ku-CPU nezidingo zememori zomthwalo wakho wokusebenza kuzokusiza ugweme ukukhokhela izinsiza ezingasetshenzisiwe.
- I-Digital Ocean ikhokhisa kancane nge-vCPU nokuningi ngenkumbulo uma kuqhathaniswa nezinye izinkundla - lokhu kungaba yisici esinqumayo sezinhlobo ezithile zokubala umthwalo wokusebenza.
*Qaphela: Ukuhlaziya kusebenzisa idatha ngenjongo evamile yokubala amanodi (inhloso-jikelele). Lezi yizimo ze-n1 GCP Compute Engine, izimo ze-m5 AWS ec2, imishini ebonakalayo ye-D2v3 Azure kanye namaconsi e-DO anama-CPU azinikele. Ngokulandelayo, kungenzeka ukwenza ucwaningo phakathi kwezinye izinhlobo zemishini ebonakalayo (eqhumayo, izinga lokungena). Uma uthi nhlΓ‘, izindleko zemishini ebonakalayo zincike ngokuqondile enanini lama-vCPU kanye nenani lenkumbulo, kodwa anginaso isiqiniseko sokuthi lokhu kucatshangelwa kuzoba yiqiniso kuzilinganiso zememori/CPU ezingajwayelekile kakhulu.
Esihlokweni , eyanyatheliswa ngo-2018, yasebenzisa iqoqo lezithenjwa elinama-cores angu-100 we-vCPU kanye nenkumbulo engu-400 GB. Ukuze uqhathanise, ngokuya ngezibalo zami, iqoqo elifanayo kwenye ngayinye yalezi zindawo (ezimweni lapho kudingeka khona) izobiza inani elilandelayo:
- I-AKS: 51465 USD/ngonyaka
- I-EKS: 43138 USD/ngonyaka
- I-GKE: 30870 USD / ngonyaka
- YENZA: 36131 USD / ngonyaka
Ngethemba ukuthi lesi sihloko kanye nencwajana yokubhalela kuzokusiza ukuthi uhlole iminikelo ephathwayo ye-Kubernetes kanye/noma wonge imali engqalasizinda yamafu ngokusebenzisa izaphulelo namanye amathuba.
I-PS evela kumhumushi
Funda futhi kubhulogi yethu:
- Β«";
- Β«";
- Β«";
- Β«".
Source: www.habr.com