I havent read/researched cilium enough yet. BPF and XDP seems to have opened a pandoras box of implementations options and have overlap with other capabilities (sec, observability, service mesh, etc.). Given Cilium today and being able to build a greenfields, money no object, kubernetes, no transaction can be lost type platform. Which parts of cilium can be used? CNI (as kube-proxy replacement?). Hubble (as observability)? Is my understanding correct, in that, Cilium CNI + Hubble brings much to the solution and really is enough/better in a CNI and operational support of the CNI (Hubble). If true, which other capabilities (sec, observability, service mesh, service discovery, sec, etc.) are best served somewhere else. Another way to ask the question is, why should I replace calico with cilium today? Not looking for a deep dive necessarily, but how do I justify switching to cilium. Thanks
I havent read/researched cilium enough yet. BPF and XDP seems to have opened a pandoras box of implementations options and have overlap with other capabilities (sec, observability, service mesh, etc.). Given Cilium today and being able to build a greenfields, money no object, kubernetes, no transaction can be lost type platform. Which parts of cilium can be used? CNI (as kube-proxy replacement?). Hubble (as observability)? Is my understanding correct, in that, Cilium CNI + Hubble brings much to the solution and really is enough/better in a CNI and operational support of the CNI (Hubble). If true, which other capabilities (sec, observability, service mesh, service discovery, sec, etc.) are best served somewhere else. Another way to ask the question is, why should I replace calico with cilium today? Not looking for a deep dive necessarily, but how do I justify switching to cilium. Thanks