Liz started off asking who is tired and a lot of hands went up, including hers 😂 #KubeCon
Liz explaining a bit of the history of service mesh and the sidecar pattern. #KubeCon
Can we move service mesh to the kernel? Mostly. L4 and below is handled but not L7 (the app layer). But Cilium already has visibility into L7. #KubeCon
Liz doing a demo showing how we can see L7 things like HTTP requests/responses/forwards. #KubeCon
Cilium uses eBPF to program the kernel and Envoy for the L7 part. #KubeCon
A CRD called CiliumEnvoyConfig configures the Envoy listeners. #KubeCon
Removing the sidecars means less containers running, and less resources used #KubeCon
These same things could be done with other service meshes (Istio, Linkerd, etc.). #KubeCon
This is pretty cool stuff. I’m super down to get rid of sidecars as much as possible. As Liz mentioned it’s a lot of duplicated resources to have one in every pod. #KubeCon
Cloud native as a paradigm involved because people wanted vendor agnostic infra, and a more declarative approach to deploying workloads. @Divya_Mohan02#KubeCon
3 categories of projects in the CNCF: Sandbox, Incubating, and Graduated. Sandbox is early with lots of innovation. Incubating projects are being used in production by more people. Graduated projects are mature and stable. #KubeCon@coffeeartgirl
During my chat last night with @LukasGentele and @fabiankramm, Lukas mentioned what a risk it was for me to join @loft_sh when I did. At that time it was mainly Lukas and Fabian and our designer (who is really rad).
My instinct at the time had actually been to join a very large company. I’d been at three early stage companies in a row and felt like I needed a break. I had never heard of Loft Labs and knew nothing about the founders.
But when I looked at the product I was very impressed. I’d been hearing people in the Kubernetes community complain about multi-tenancy pain for years. Virtual clusters were such a new approach and very smart I thought. (At that point they were in the commercial product only.)